Compare commits

..

1 Commits

Author SHA1 Message Date
Weny Xu
78195fac6e Revert "fix: align backend_tls default value with example config (#7496)"
This reverts commit 83932c8c9e.
2025-12-31 14:47:59 +08:00
62 changed files with 454 additions and 2515 deletions

View File

@@ -70,23 +70,19 @@ runs:
--wait \
--wait-for-jobs
- name: Wait for GreptimeDB
uses: nick-fields/retry@v3
with:
timeout_minutes: 3
max_attempts: 1
shell: bash
command: |
while true; do
PHASE=$(kubectl -n my-greptimedb get gtc my-greptimedb -o jsonpath='{.status.clusterPhase}')
if [ "$PHASE" == "Running" ]; then
echo "Cluster is ready"
break
else
echo "Cluster is not ready yet: Current phase: $PHASE"
kubectl get pods -n my-greptimedb
sleep 5 # wait for 5 seconds before check again.
fi
done
shell: bash
run: |
while true; do
PHASE=$(kubectl -n my-greptimedb get gtc my-greptimedb -o jsonpath='{.status.clusterPhase}')
if [ "$PHASE" == "Running" ]; then
echo "Cluster is ready"
break
else
echo "Cluster is not ready yet: Current phase: $PHASE"
kubectl get pods -n my-greptimedb
sleep 5 # wait for 5 seconds before check again.
fi
done
- name: Print GreptimeDB info
if: always()
shell: bash

3
Cargo.lock generated
View File

@@ -4062,7 +4062,6 @@ dependencies = [
"mito2",
"num_cpus",
"object-store",
"partition",
"prometheus",
"prost 0.13.5",
"query",
@@ -4100,7 +4099,6 @@ dependencies = [
"num-traits",
"ordered-float 4.6.0",
"paste",
"regex",
"serde",
"serde_json",
"snafu 0.8.6",
@@ -9476,7 +9474,6 @@ dependencies = [
"ahash 0.8.12",
"api",
"arrow",
"arrow-schema",
"async-trait",
"catalog",
"chrono",

View File

@@ -895,7 +895,7 @@ pub fn is_column_type_value_eq(
.unwrap_or(false)
}
pub fn encode_json_value(value: JsonValue) -> v1::JsonValue {
fn encode_json_value(value: JsonValue) -> v1::JsonValue {
fn helper(json: JsonVariant) -> v1::JsonValue {
let value = match json {
JsonVariant::Null => None,

View File

@@ -17,8 +17,8 @@ use std::collections::HashMap;
use arrow_schema::extension::{EXTENSION_TYPE_METADATA_KEY, EXTENSION_TYPE_NAME_KEY};
use datatypes::schema::{
COMMENT_KEY, ColumnDefaultConstraint, ColumnSchema, FULLTEXT_KEY, FulltextAnalyzer,
FulltextBackend, FulltextOptions, INVERTED_INDEX_KEY, Metadata, SKIPPING_INDEX_KEY,
SkippingIndexOptions, SkippingIndexType,
FulltextBackend, FulltextOptions, INVERTED_INDEX_KEY, SKIPPING_INDEX_KEY, SkippingIndexOptions,
SkippingIndexType,
};
use greptime_proto::v1::{
Analyzer, FulltextBackend as PbFulltextBackend, SkippingIndexType as PbSkippingIndexType,
@@ -36,14 +36,6 @@ const INVERTED_INDEX_GRPC_KEY: &str = "inverted_index";
/// Key used to store skip index options in gRPC column options.
const SKIPPING_INDEX_GRPC_KEY: &str = "skipping_index";
const COLUMN_OPTION_MAPPINGS: [(&str, &str); 5] = [
(FULLTEXT_GRPC_KEY, FULLTEXT_KEY),
(INVERTED_INDEX_GRPC_KEY, INVERTED_INDEX_KEY),
(SKIPPING_INDEX_GRPC_KEY, SKIPPING_INDEX_KEY),
(EXTENSION_TYPE_NAME_KEY, EXTENSION_TYPE_NAME_KEY),
(EXTENSION_TYPE_METADATA_KEY, EXTENSION_TYPE_METADATA_KEY),
];
/// Tries to construct a `ColumnSchema` from the given `ColumnDef`.
pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
let data_type = ColumnDataTypeWrapper::try_new(
@@ -139,21 +131,6 @@ pub fn try_as_column_def(column_schema: &ColumnSchema, is_primary_key: bool) ->
})
}
/// Collect the [ColumnOptions] into the [Metadata] that can be used in, for example, [ColumnSchema].
pub fn collect_column_options(column_options: Option<&ColumnOptions>) -> Metadata {
let Some(ColumnOptions { options }) = column_options else {
return Metadata::default();
};
let mut metadata = Metadata::with_capacity(options.len());
for (x, y) in COLUMN_OPTION_MAPPINGS {
if let Some(v) = options.get(x) {
metadata.insert(y.to_string(), v.clone());
}
}
metadata
}
/// Constructs a `ColumnOptions` from the given `ColumnSchema`.
pub fn options_from_column_schema(column_schema: &ColumnSchema) -> Option<ColumnOptions> {
let mut options = ColumnOptions::default();

View File

@@ -399,8 +399,8 @@ impl InformationSchemaColumnsBuilder {
self.is_nullables.push(Some("No"));
}
self.column_types.push(Some(&data_type));
let column_comment = column_schema.column_comment().map(|x| x.as_ref());
self.column_comments.push(column_comment);
self.column_comments
.push(column_schema.column_comment().map(|x| x.as_ref()));
}
fn finish(&mut self) -> Result<RecordBatch> {

View File

@@ -92,7 +92,7 @@ impl StoreConfig {
pub fn tls_config(&self) -> Option<TlsOption> {
if self.backend_tls_mode != TlsMode::Disable {
Some(TlsOption {
mode: self.backend_tls_mode,
mode: self.backend_tls_mode.clone(),
cert_path: self.backend_tls_cert_path.clone(),
key_path: self.backend_tls_key_path.clone(),
ca_cert_path: self.backend_tls_ca_cert_path.clone(),

View File

@@ -236,7 +236,7 @@ impl StartCommand {
};
let tls_opts = TlsOption::new(
self.tls_mode,
self.tls_mode.clone(),
self.tls_cert_path.clone(),
self.tls_key_path.clone(),
self.tls_watch,

View File

@@ -261,7 +261,7 @@ impl StartCommand {
};
let tls_opts = TlsOption::new(
self.tls_mode,
self.tls_mode.clone(),
self.tls_cert_path.clone(),
self.tls_key_path.clone(),
self.tls_watch,

View File

@@ -530,49 +530,6 @@ impl Display for EnterStagingRegion {
}
}
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
pub struct RemapManifest {
pub region_id: RegionId,
/// Regions to remap manifests from.
pub input_regions: Vec<RegionId>,
/// For each old region, which new regions should receive its files
pub region_mapping: HashMap<RegionId, Vec<RegionId>>,
/// New partition expressions for the new regions.
pub new_partition_exprs: HashMap<RegionId, String>,
}
impl Display for RemapManifest {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
write!(
f,
"RemapManifest(region_id={}, input_regions={:?}, region_mapping={:?}, new_partition_exprs={:?})",
self.region_id, self.input_regions, self.region_mapping, self.new_partition_exprs
)
}
}
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
pub struct ApplyStagingManifest {
/// The region ID to apply the staging manifest to.
pub region_id: RegionId,
/// The partition expression of the staging region.
pub partition_expr: String,
/// The region that stores the staging manifests in its staging blob storage.
pub central_region_id: RegionId,
/// The relative path to the staging manifest within the central region's staging blob storage.
pub manifest_path: String,
}
impl Display for ApplyStagingManifest {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
write!(
f,
"ApplyStagingManifest(region_id={}, partition_expr={}, central_region_id={}, manifest_path={})",
self.region_id, self.partition_expr, self.central_region_id, self.manifest_path
)
}
}
#[derive(Debug, Clone, Serialize, Deserialize, Display, PartialEq)]
pub enum Instruction {
/// Opens regions.
@@ -602,10 +559,6 @@ pub enum Instruction {
Suspend,
/// Makes regions enter staging state.
EnterStagingRegions(Vec<EnterStagingRegion>),
/// Remaps manifests for a region.
RemapManifest(RemapManifest),
/// Applies staging manifests for a region.
ApplyStagingManifests(Vec<ApplyStagingManifest>),
}
impl Instruction {
@@ -784,48 +737,6 @@ impl EnterStagingRegionsReply {
}
}
#[derive(Debug, Serialize, Deserialize, PartialEq, Eq, Clone)]
pub struct RemapManifestReply {
/// Returns false if the region does not exist.
pub exists: bool,
/// A map from region IDs to their corresponding remapped manifest paths.
pub manifest_paths: HashMap<RegionId, String>,
/// Return error if any during the operation.
pub error: Option<String>,
}
impl Display for RemapManifestReply {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
write!(
f,
"RemapManifestReply(manifest_paths={:?}, error={:?})",
self.manifest_paths, self.error
)
}
}
#[derive(Debug, Serialize, Deserialize, PartialEq, Eq, Clone)]
pub struct ApplyStagingManifestsReply {
pub replies: Vec<ApplyStagingManifestReply>,
}
impl ApplyStagingManifestsReply {
pub fn new(replies: Vec<ApplyStagingManifestReply>) -> Self {
Self { replies }
}
}
#[derive(Debug, Serialize, Deserialize, PartialEq, Eq, Clone)]
pub struct ApplyStagingManifestReply {
pub region_id: RegionId,
/// Returns true if the region is ready to serve reads and writes.
pub ready: bool,
/// Indicates whether the region exists.
pub exists: bool,
/// Return error if any during the operation.
pub error: Option<String>,
}
#[derive(Debug, Serialize, Deserialize, PartialEq, Eq, Clone)]
#[serde(tag = "type", rename_all = "snake_case")]
pub enum InstructionReply {
@@ -847,8 +758,6 @@ pub enum InstructionReply {
GetFileRefs(GetFileRefsReply),
GcRegions(GcRegionsReply),
EnterStagingRegions(EnterStagingRegionsReply),
RemapManifest(RemapManifestReply),
ApplyStagingManifests(ApplyStagingManifestsReply),
}
impl Display for InstructionReply {
@@ -872,12 +781,6 @@ impl Display for InstructionReply {
reply.replies
)
}
Self::RemapManifest(reply) => write!(f, "InstructionReply::RemapManifest({})", reply),
Self::ApplyStagingManifests(reply) => write!(
f,
"InstructionReply::ApplyStagingManifests({:?})",
reply.replies
),
}
}
}
@@ -925,20 +828,6 @@ impl InstructionReply {
_ => panic!("Expected EnterStagingRegion reply"),
}
}
pub fn expect_remap_manifest_reply(self) -> RemapManifestReply {
match self {
Self::RemapManifest(reply) => reply,
_ => panic!("Expected RemapManifest reply"),
}
}
pub fn expect_apply_staging_manifests_reply(self) -> Vec<ApplyStagingManifestReply> {
match self {
Self::ApplyStagingManifests(reply) => reply.replies,
_ => panic!("Expected ApplyStagingManifest reply"),
}
}
}
#[cfg(test)]

View File

@@ -868,8 +868,6 @@ impl PgStore {
let client = match pool.get().await {
Ok(client) => client,
Err(e) => {
// We need to log the debug for the error to help diagnose the issue.
common_telemetry::error!(e; "Failed to get Postgres connection.");
return GetPostgresConnectionSnafu {
reason: e.to_string(),
}

View File

@@ -77,5 +77,4 @@ common-query.workspace = true
common-test-util.workspace = true
datafusion-common.workspace = true
mito2 = { workspace = true, features = ["test"] }
partition.workspace = true
session.workspace = true

View File

@@ -201,7 +201,6 @@ pub enum Error {
ShutdownServer {
#[snafu(implicit)]
location: Location,
#[snafu(source)]
source: servers::error::Error,
},
@@ -209,7 +208,6 @@ pub enum Error {
ShutdownInstance {
#[snafu(implicit)]
location: Location,
#[snafu(source)]
source: BoxedError,
},

View File

@@ -22,7 +22,6 @@ use common_telemetry::error;
use snafu::OptionExt;
use store_api::storage::GcReport;
mod apply_staging_manifest;
mod close_region;
mod downgrade_region;
mod enter_staging;
@@ -30,10 +29,8 @@ mod file_ref;
mod flush_region;
mod gc_worker;
mod open_region;
mod remap_manifest;
mod upgrade_region;
use crate::heartbeat::handler::apply_staging_manifest::ApplyStagingManifestsHandler;
use crate::heartbeat::handler::close_region::CloseRegionsHandler;
use crate::heartbeat::handler::downgrade_region::DowngradeRegionsHandler;
use crate::heartbeat::handler::enter_staging::EnterStagingRegionsHandler;
@@ -41,7 +38,6 @@ use crate::heartbeat::handler::file_ref::GetFileRefsHandler;
use crate::heartbeat::handler::flush_region::FlushRegionsHandler;
use crate::heartbeat::handler::gc_worker::GcRegionsHandler;
use crate::heartbeat::handler::open_region::OpenRegionsHandler;
use crate::heartbeat::handler::remap_manifest::RemapManifestHandler;
use crate::heartbeat::handler::upgrade_region::UpgradeRegionsHandler;
use crate::heartbeat::task_tracker::TaskTracker;
use crate::region_server::RegionServer;
@@ -132,10 +128,6 @@ impl RegionHeartbeatResponseHandler {
Instruction::EnterStagingRegions(_) => {
Ok(Some(Box::new(EnterStagingRegionsHandler.into())))
}
Instruction::RemapManifest(_) => Ok(Some(Box::new(RemapManifestHandler.into()))),
Instruction::ApplyStagingManifests(_) => {
Ok(Some(Box::new(ApplyStagingManifestsHandler.into())))
}
}
}
}
@@ -150,8 +142,6 @@ pub enum InstructionHandlers {
GetFileRefs(GetFileRefsHandler),
GcRegions(GcRegionsHandler),
EnterStagingRegions(EnterStagingRegionsHandler),
RemapManifest(RemapManifestHandler),
ApplyStagingManifests(ApplyStagingManifestsHandler),
}
macro_rules! impl_from_handler {
@@ -174,9 +164,7 @@ impl_from_handler!(
UpgradeRegionsHandler => UpgradeRegions,
GetFileRefsHandler => GetFileRefs,
GcRegionsHandler => GcRegions,
EnterStagingRegionsHandler => EnterStagingRegions,
RemapManifestHandler => RemapManifest,
ApplyStagingManifestsHandler => ApplyStagingManifests
EnterStagingRegionsHandler => EnterStagingRegions
);
macro_rules! dispatch_instr {
@@ -221,9 +209,7 @@ dispatch_instr!(
UpgradeRegions => UpgradeRegions,
GetFileRefs => GetFileRefs,
GcRegions => GcRegions,
EnterStagingRegions => EnterStagingRegions,
RemapManifest => RemapManifest,
ApplyStagingManifests => ApplyStagingManifests,
EnterStagingRegions => EnterStagingRegions
);
#[async_trait]

View File

@@ -1,287 +0,0 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use common_meta::instruction::{
ApplyStagingManifest, ApplyStagingManifestReply, ApplyStagingManifestsReply, InstructionReply,
};
use common_telemetry::{error, warn};
use futures::future::join_all;
use store_api::region_request::{ApplyStagingManifestRequest, RegionRequest};
use crate::heartbeat::handler::{HandlerContext, InstructionHandler};
pub struct ApplyStagingManifestsHandler;
#[async_trait::async_trait]
impl InstructionHandler for ApplyStagingManifestsHandler {
type Instruction = Vec<ApplyStagingManifest>;
async fn handle(
&self,
ctx: &HandlerContext,
requests: Self::Instruction,
) -> Option<InstructionReply> {
let results = join_all(
requests
.into_iter()
.map(|request| Self::handle_apply_staging_manifest(ctx, request)),
)
.await;
Some(InstructionReply::ApplyStagingManifests(
ApplyStagingManifestsReply::new(results),
))
}
}
impl ApplyStagingManifestsHandler {
async fn handle_apply_staging_manifest(
ctx: &HandlerContext,
request: ApplyStagingManifest,
) -> ApplyStagingManifestReply {
let Some(leader) = ctx.region_server.is_region_leader(request.region_id) else {
warn!("Region: {} is not found", request.region_id);
return ApplyStagingManifestReply {
region_id: request.region_id,
exists: false,
ready: false,
error: None,
};
};
if !leader {
warn!("Region: {} is not leader", request.region_id);
return ApplyStagingManifestReply {
region_id: request.region_id,
exists: true,
ready: false,
error: Some("Region is not leader".into()),
};
}
match ctx
.region_server
.handle_request(
request.region_id,
RegionRequest::ApplyStagingManifest(ApplyStagingManifestRequest {
partition_expr: request.partition_expr,
central_region_id: request.central_region_id,
manifest_path: request.manifest_path,
}),
)
.await
{
Ok(_) => ApplyStagingManifestReply {
region_id: request.region_id,
exists: true,
ready: true,
error: None,
},
Err(err) => {
error!(err; "Failed to apply staging manifest");
ApplyStagingManifestReply {
region_id: request.region_id,
exists: true,
ready: false,
error: Some(format!("{err:?}")),
}
}
}
}
}
#[cfg(test)]
mod tests {
use std::collections::HashMap;
use std::sync::Arc;
use common_meta::instruction::RemapManifest;
use datatypes::value::Value;
use mito2::config::MitoConfig;
use mito2::engine::MITO_ENGINE_NAME;
use mito2::test_util::{CreateRequestBuilder, TestEnv};
use partition::expr::{PartitionExpr, col};
use store_api::path_utils::table_dir;
use store_api::region_engine::RegionRole;
use store_api::region_request::EnterStagingRequest;
use store_api::storage::RegionId;
use super::*;
use crate::heartbeat::handler::remap_manifest::RemapManifestHandler;
use crate::region_server::RegionServer;
use crate::tests::{MockRegionEngine, mock_region_server};
#[tokio::test]
async fn test_region_not_exist() {
let mut mock_region_server = mock_region_server();
let (mock_engine, _) = MockRegionEngine::new(MITO_ENGINE_NAME);
mock_region_server.register_engine(mock_engine);
let handler_context = HandlerContext::new_for_test(mock_region_server);
let region_id = RegionId::new(1024, 1);
let reply = ApplyStagingManifestsHandler
.handle(
&handler_context,
vec![ApplyStagingManifest {
region_id,
partition_expr: "".to_string(),
central_region_id: RegionId::new(1024, 9999), // use a dummy value
manifest_path: "".to_string(),
}],
)
.await
.unwrap();
let replies = reply.expect_apply_staging_manifests_reply();
let reply = &replies[0];
assert!(!reply.exists);
assert!(!reply.ready);
assert!(reply.error.is_none());
}
#[tokio::test]
async fn test_region_not_leader() {
let mock_region_server = mock_region_server();
let region_id = RegionId::new(1024, 1);
let (mock_engine, _) =
MockRegionEngine::with_custom_apply_fn(MITO_ENGINE_NAME, |region_engine| {
region_engine.mock_role = Some(Some(RegionRole::Follower));
region_engine.handle_request_mock_fn = Some(Box::new(|_, _| Ok(0)));
});
mock_region_server.register_test_region(region_id, mock_engine);
let handler_context = HandlerContext::new_for_test(mock_region_server);
let region_id = RegionId::new(1024, 1);
let reply = ApplyStagingManifestsHandler
.handle(
&handler_context,
vec![ApplyStagingManifest {
region_id,
partition_expr: "".to_string(),
central_region_id: RegionId::new(1024, 2),
manifest_path: "".to_string(),
}],
)
.await
.unwrap();
let replies = reply.expect_apply_staging_manifests_reply();
let reply = &replies[0];
assert!(reply.exists);
assert!(!reply.ready);
assert!(reply.error.is_some());
}
fn range_expr(col_name: &str, start: i64, end: i64) -> PartitionExpr {
col(col_name)
.gt_eq(Value::Int64(start))
.and(col(col_name).lt(Value::Int64(end)))
}
async fn prepare_region(region_server: &RegionServer) {
let region_specs = [
(RegionId::new(1024, 1), range_expr("x", 0, 49)),
(RegionId::new(1024, 2), range_expr("x", 49, 100)),
];
for (region_id, partition_expr) in region_specs {
let builder = CreateRequestBuilder::new();
let mut create_req = builder.build();
create_req.table_dir = table_dir("test", 1024);
region_server
.handle_request(region_id, RegionRequest::Create(create_req))
.await
.unwrap();
region_server
.handle_request(
region_id,
RegionRequest::EnterStaging(EnterStagingRequest {
partition_expr: partition_expr.as_json_str().unwrap(),
}),
)
.await
.unwrap();
}
}
#[tokio::test]
async fn test_apply_staging_manifest() {
common_telemetry::init_default_ut_logging();
let mut region_server = mock_region_server();
let region_id = RegionId::new(1024, 1);
let mut engine_env = TestEnv::new().await;
let engine = engine_env.create_engine(MitoConfig::default()).await;
region_server.register_engine(Arc::new(engine.clone()));
prepare_region(&region_server).await;
let handler_context = HandlerContext::new_for_test(region_server);
let region_id2 = RegionId::new(1024, 2);
let reply = RemapManifestHandler
.handle(
&handler_context,
RemapManifest {
region_id,
input_regions: vec![region_id, region_id2],
region_mapping: HashMap::from([
// [0,49) <- [0, 50)
(region_id, vec![region_id]),
// [49, 100) <- [0, 50), [50,100)
(region_id2, vec![region_id, region_id2]),
]),
new_partition_exprs: HashMap::from([
(region_id, range_expr("x", 0, 49).as_json_str().unwrap()),
(region_id2, range_expr("x", 49, 100).as_json_str().unwrap()),
]),
},
)
.await
.unwrap();
let reply = reply.expect_remap_manifest_reply();
assert!(reply.exists);
assert!(reply.error.is_none(), "{}", reply.error.unwrap());
assert_eq!(reply.manifest_paths.len(), 2);
let manifest_path_1 = reply.manifest_paths[&region_id].clone();
let manifest_path_2 = reply.manifest_paths[&region_id2].clone();
let reply = ApplyStagingManifestsHandler
.handle(
&handler_context,
vec![ApplyStagingManifest {
region_id,
partition_expr: range_expr("x", 0, 49).as_json_str().unwrap(),
central_region_id: region_id,
manifest_path: manifest_path_1,
}],
)
.await
.unwrap();
let replies = reply.expect_apply_staging_manifests_reply();
let reply = &replies[0];
assert!(reply.exists);
assert!(reply.ready);
assert!(reply.error.is_none());
// partition expr mismatch
let reply = ApplyStagingManifestsHandler
.handle(
&handler_context,
vec![ApplyStagingManifest {
region_id: region_id2,
partition_expr: range_expr("x", 50, 100).as_json_str().unwrap(),
central_region_id: region_id,
manifest_path: manifest_path_2,
}],
)
.await
.unwrap();
let replies = reply.expect_apply_staging_manifests_reply();
let reply = &replies[0];
assert!(reply.exists);
assert!(!reply.ready);
assert!(reply.error.is_some());
}
}

View File

@@ -1,246 +0,0 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use common_meta::instruction::{InstructionReply, RemapManifest, RemapManifestReply};
use common_telemetry::warn;
use store_api::region_engine::RemapManifestsRequest;
use crate::heartbeat::handler::{HandlerContext, InstructionHandler};
pub struct RemapManifestHandler;
#[async_trait::async_trait]
impl InstructionHandler for RemapManifestHandler {
type Instruction = RemapManifest;
async fn handle(
&self,
ctx: &HandlerContext,
request: Self::Instruction,
) -> Option<InstructionReply> {
let RemapManifest {
region_id,
input_regions,
region_mapping,
new_partition_exprs,
} = request;
let Some(leader) = ctx.region_server.is_region_leader(region_id) else {
warn!("Region: {} is not found", region_id);
return Some(InstructionReply::RemapManifest(RemapManifestReply {
exists: false,
manifest_paths: Default::default(),
error: None,
}));
};
if !leader {
warn!("Region: {} is not leader", region_id);
return Some(InstructionReply::RemapManifest(RemapManifestReply {
exists: true,
manifest_paths: Default::default(),
error: Some("Region is not leader".into()),
}));
}
let reply = match ctx
.region_server
.remap_manifests(RemapManifestsRequest {
region_id,
input_regions,
region_mapping,
new_partition_exprs,
})
.await
{
Ok(result) => InstructionReply::RemapManifest(RemapManifestReply {
exists: true,
manifest_paths: result.manifest_paths,
error: None,
}),
Err(e) => InstructionReply::RemapManifest(RemapManifestReply {
exists: true,
manifest_paths: Default::default(),
error: Some(format!("{e:?}")),
}),
};
Some(reply)
}
}
#[cfg(test)]
mod tests {
use std::collections::HashMap;
use std::sync::Arc;
use common_meta::instruction::RemapManifest;
use datatypes::value::Value;
use mito2::config::MitoConfig;
use mito2::engine::MITO_ENGINE_NAME;
use mito2::test_util::{CreateRequestBuilder, TestEnv};
use partition::expr::{PartitionExpr, col};
use store_api::path_utils::table_dir;
use store_api::region_engine::RegionRole;
use store_api::region_request::{EnterStagingRequest, RegionRequest};
use store_api::storage::RegionId;
use crate::heartbeat::handler::remap_manifest::RemapManifestHandler;
use crate::heartbeat::handler::{HandlerContext, InstructionHandler};
use crate::region_server::RegionServer;
use crate::tests::{MockRegionEngine, mock_region_server};
#[tokio::test]
async fn test_region_not_exist() {
let mut mock_region_server = mock_region_server();
let (mock_engine, _) = MockRegionEngine::new(MITO_ENGINE_NAME);
mock_region_server.register_engine(mock_engine);
let handler_context = HandlerContext::new_for_test(mock_region_server);
let region_id = RegionId::new(1024, 1);
let reply = RemapManifestHandler
.handle(
&handler_context,
RemapManifest {
region_id,
input_regions: vec![],
region_mapping: HashMap::new(),
new_partition_exprs: HashMap::new(),
},
)
.await
.unwrap();
let reply = &reply.expect_remap_manifest_reply();
assert!(!reply.exists);
assert!(reply.error.is_none());
assert!(reply.manifest_paths.is_empty());
}
#[tokio::test]
async fn test_region_not_leader() {
let mock_region_server = mock_region_server();
let region_id = RegionId::new(1024, 1);
let (mock_engine, _) =
MockRegionEngine::with_custom_apply_fn(MITO_ENGINE_NAME, |region_engine| {
region_engine.mock_role = Some(Some(RegionRole::Follower));
region_engine.handle_request_mock_fn = Some(Box::new(|_, _| Ok(0)));
});
mock_region_server.register_test_region(region_id, mock_engine);
let handler_context = HandlerContext::new_for_test(mock_region_server);
let reply = RemapManifestHandler
.handle(
&handler_context,
RemapManifest {
region_id,
input_regions: vec![],
region_mapping: HashMap::new(),
new_partition_exprs: HashMap::new(),
},
)
.await
.unwrap();
let reply = reply.expect_remap_manifest_reply();
assert!(reply.exists);
assert!(reply.error.is_some());
}
fn range_expr(col_name: &str, start: i64, end: i64) -> PartitionExpr {
col(col_name)
.gt_eq(Value::Int64(start))
.and(col(col_name).lt(Value::Int64(end)))
}
async fn prepare_region(region_server: &RegionServer) {
let region_specs = [
(RegionId::new(1024, 1), range_expr("x", 0, 50)),
(RegionId::new(1024, 2), range_expr("x", 50, 100)),
];
for (region_id, partition_expr) in region_specs {
let builder = CreateRequestBuilder::new();
let mut create_req = builder.build();
create_req.table_dir = table_dir("test", 1024);
region_server
.handle_request(region_id, RegionRequest::Create(create_req))
.await
.unwrap();
region_server
.handle_request(
region_id,
RegionRequest::EnterStaging(EnterStagingRequest {
partition_expr: partition_expr.as_json_str().unwrap(),
}),
)
.await
.unwrap();
}
}
#[tokio::test]
async fn test_remap_manifest() {
common_telemetry::init_default_ut_logging();
let mut region_server = mock_region_server();
let region_id = RegionId::new(1024, 1);
let mut engine_env = TestEnv::new().await;
let engine = engine_env.create_engine(MitoConfig::default()).await;
region_server.register_engine(Arc::new(engine.clone()));
prepare_region(&region_server).await;
let handler_context = HandlerContext::new_for_test(region_server);
let region_id2 = RegionId::new(1024, 2);
let reply = RemapManifestHandler
.handle(
&handler_context,
RemapManifest {
region_id,
input_regions: vec![region_id, region_id2],
region_mapping: HashMap::from([
(region_id, vec![region_id]),
(region_id2, vec![region_id]),
]),
new_partition_exprs: HashMap::from([(
region_id,
range_expr("x", 0, 100).as_json_str().unwrap(),
)]),
},
)
.await
.unwrap();
let reply = reply.expect_remap_manifest_reply();
assert!(reply.exists);
assert!(reply.error.is_none(), "{}", reply.error.unwrap());
assert_eq!(reply.manifest_paths.len(), 1);
// Remap failed
let reply = RemapManifestHandler
.handle(
&handler_context,
RemapManifest {
region_id,
input_regions: vec![region_id],
region_mapping: HashMap::from([
(region_id, vec![region_id]),
(region_id2, vec![region_id]),
]),
new_partition_exprs: HashMap::from([(
region_id,
range_expr("x", 0, 100).as_json_str().unwrap(),
)]),
},
)
.await
.unwrap();
let reply = reply.expect_remap_manifest_reply();
assert!(reply.exists);
assert!(reply.error.is_some());
assert!(reply.manifest_paths.is_empty());
}
}

View File

@@ -65,9 +65,8 @@ use store_api::metric_engine_consts::{
FILE_ENGINE_NAME, LOGICAL_TABLE_METADATA_KEY, METRIC_ENGINE_NAME,
};
use store_api::region_engine::{
RegionEngineRef, RegionManifestInfo, RegionRole, RegionStatistic, RemapManifestsRequest,
RemapManifestsResponse, SetRegionRoleStateResponse, SettableRegionRoleState,
SyncRegionFromRequest,
RegionEngineRef, RegionManifestInfo, RegionRole, RegionStatistic, SetRegionRoleStateResponse,
SettableRegionRoleState, SyncRegionFromRequest,
};
use store_api::region_request::{
AffectedRows, BatchRegionDdlRequest, RegionCatchupRequest, RegionCloseRequest,
@@ -605,25 +604,6 @@ impl RegionServer {
.await
}
/// Remaps manifests from old regions to new regions.
pub async fn remap_manifests(
&self,
request: RemapManifestsRequest,
) -> Result<RemapManifestsResponse> {
let region_id = request.region_id;
let engine_with_status = self
.inner
.region_map
.get(&region_id)
.with_context(|| RegionNotFoundSnafu { region_id })?;
engine_with_status
.engine()
.remap_manifests(request)
.await
.with_context(|_| HandleRegionRequestSnafu { region_id })
}
fn is_suspended(&self) -> bool {
self.suspend.load(Ordering::Relaxed)
}

View File

@@ -28,7 +28,6 @@ num = "0.4"
num-traits = "0.2"
ordered-float.workspace = true
paste.workspace = true
regex.workspace = true
serde.workspace = true
serde_json.workspace = true
snafu.workspace = true

View File

@@ -816,7 +816,7 @@ mod tests {
let result = encode_by_struct(&json_struct, json);
assert_eq!(
result.unwrap_err().to_string(),
r#"Cannot cast value bar to "<Number>""#
"Cannot cast value bar to Number(I64)"
);
let json = json!({

View File

@@ -13,13 +13,12 @@
// limitations under the License.
use std::collections::BTreeMap;
use std::fmt::{Debug, Display, Formatter};
use std::fmt::{Display, Formatter};
use std::str::FromStr;
use std::sync::{Arc, LazyLock};
use std::sync::Arc;
use arrow::datatypes::DataType as ArrowDataType;
use common_base::bytes::Bytes;
use regex::{Captures, Regex};
use serde::{Deserialize, Serialize};
use snafu::ResultExt;
@@ -135,24 +134,24 @@ impl From<&ConcreteDataType> for JsonNativeType {
impl Display for JsonNativeType {
fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
match self {
JsonNativeType::Null => write!(f, r#""<Null>""#),
JsonNativeType::Bool => write!(f, r#""<Bool>""#),
JsonNativeType::Number(_) => {
write!(f, r#""<Number>""#)
JsonNativeType::Null => write!(f, "Null"),
JsonNativeType::Bool => write!(f, "Bool"),
JsonNativeType::Number(t) => {
write!(f, "Number({t:?})")
}
JsonNativeType::String => write!(f, r#""<String>""#),
JsonNativeType::String => write!(f, "String"),
JsonNativeType::Array(item_type) => {
write!(f, "[{}]", item_type)
write!(f, "Array[{}]", item_type)
}
JsonNativeType::Object(object) => {
write!(
f,
"{{{}}}",
"Object{{{}}}",
object
.iter()
.map(|(k, v)| format!(r#""{k}":{v}"#))
.map(|(k, v)| format!(r#""{k}": {v}"#))
.collect::<Vec<_>>()
.join(",")
.join(", ")
)
}
}
@@ -184,11 +183,7 @@ impl JsonType {
}
}
pub fn is_native_type(&self) -> bool {
matches!(self.format, JsonFormat::Native(_))
}
pub fn native_type(&self) -> &JsonNativeType {
pub(crate) fn native_type(&self) -> &JsonNativeType {
match &self.format {
JsonFormat::Jsonb => &JsonNativeType::String,
JsonFormat::Native(x) => x.as_ref(),
@@ -402,81 +397,8 @@ pub fn jsonb_to_string(val: &[u8]) -> Result<String> {
/// Converts a json type value to serde_json::Value
pub fn jsonb_to_serde_json(val: &[u8]) -> Result<serde_json::Value> {
let json_string = jsonb_to_string(val)?;
jsonb_string_to_serde_value(&json_string)
}
/// Attempts to deserialize a JSON text into `serde_json::Value`, with a best-effort
/// fallback for Rust-style Unicode escape sequences.
///
/// This function is intended to be used on JSON strings produced from the internal
/// JSONB representation (e.g. via [`jsonb_to_string`]). It first calls
/// `serde_json::Value::from_str` directly. If that succeeds, the parsed value is
/// returned as-is.
///
/// If the initial parse fails, the input is scanned for Rust-style Unicode code
/// point escapes of the form `\\u{H...}` (a backslash, `u`, an opening brace,
/// followed by 16 hexadecimal digits, and a closing brace). Each such escape is
/// converted into JSON-compatible UTF16 escape sequences:
///
/// - For code points in the Basic Multilingual Plane (≤ `0xFFFF`), the escape is
/// converted to a single JSON `\\uXXXX` sequence with four uppercase hex digits.
/// - For code points above `0xFFFF` and less than Unicode max code point `0x10FFFF`,
/// the code point is encoded as a UTF16 surrogate pair and emitted as two consecutive
/// `\\uXXXX` sequences (as JSON format required).
///
/// After this normalization, the function retries parsing the resulting string as
/// JSON and returns the deserialized value or a `DeserializeSnafu` error if it
/// still cannot be parsed.
fn jsonb_string_to_serde_value(json: &str) -> Result<serde_json::Value> {
match serde_json::Value::from_str(json) {
Ok(v) => Ok(v),
Err(e) => {
// If above deserialization is failed, the JSON string might contain some Rust chars
// that are somehow incorrectly represented as Unicode code point literal. For example,
// "\u{fe0f}". We have to convert them to JSON compatible format, like "\uFE0F", then
// try to deserialize the JSON string again.
if !e.is_syntax() || !e.to_string().contains("invalid escape") {
return Err(e).context(DeserializeSnafu { json });
}
static UNICODE_CODE_POINT_PATTERN: LazyLock<Regex> = LazyLock::new(|| {
// Match literal "\u{...}" sequences, capturing 16 (code point range) hex digits
// inside braces.
Regex::new(r"\\u\{([0-9a-fA-F]{1,6})}").unwrap_or_else(|e| panic!("{}", e))
});
let v = UNICODE_CODE_POINT_PATTERN.replace_all(json, |caps: &Captures| {
// Extract the hex payload (without braces) and parse to a code point.
let hex = &caps[1];
let Ok(code) = u32::from_str_radix(hex, 16) else {
// On parse failure, leave the original escape sequence unchanged.
return caps[0].to_string();
};
if code <= 0xFFFF {
// Basic Multilingual Plane: JSON can represent this directly as \uXXXX.
format!("\\u{:04X}", code)
} else if code > 0x10FFFF {
// Beyond max Unicode code point
caps[0].to_string()
} else {
// Supplementary planes: JSON needs UTF-16 surrogate pairs.
// Convert the code point to a 20-bit value.
let code = code - 0x10000;
// High surrogate: top 10 bits, offset by 0xD800.
let high = 0xD800 + ((code >> 10) & 0x3FF);
// Low surrogate: bottom 10 bits, offset by 0xDC00.
let low = 0xDC00 + (code & 0x3FF);
// Emit two \uXXXX escapes in sequence.
format!("\\u{:04X}\\u{:04X}", high, low)
}
});
serde_json::Value::from_str(&v).context(DeserializeSnafu { json })
}
}
serde_json::Value::from_str(json_string.as_str())
.context(DeserializeSnafu { json: json_string })
}
/// Parses a string to a json type value
@@ -491,49 +413,6 @@ mod tests {
use super::*;
use crate::json::JsonStructureSettings;
#[test]
fn test_jsonb_string_to_serde_value() -> Result<()> {
let valid_cases = vec![
(r#"{"data": "simple ascii"}"#, r#"{"data":"simple ascii"}"#),
(
r#"{"data": "Greek sigma: \u{03a3}"}"#,
r#"{"data":"Greek sigma: Σ"}"#,
),
(
r#"{"data": "Joker card: \u{1f0df}"}"#,
r#"{"data":"Joker card: 🃟"}"#,
),
(
r#"{"data": "BMP boundary: \u{ffff}"}"#,
r#"{"data":"BMP boundary: ￿"}"#,
),
(
r#"{"data": "Supplementary min: \u{10000}"}"#,
r#"{"data":"Supplementary min: 𐀀"}"#,
),
(
r#"{"data": "Supplementary max: \u{10ffff}"}"#,
r#"{"data":"Supplementary max: 􏿿"}"#,
),
];
for (input, expect) in valid_cases {
let v = jsonb_string_to_serde_value(input)?;
assert_eq!(v.to_string(), expect);
}
let invalid_cases = vec![
r#"{"data": "Invalid hex: \u{gggg}"}"#,
r#"{"data": "Beyond max Unicode code point: \u{110000}"}"#,
r#"{"data": "Out of range: \u{1100000}"}"#, // 7 digit
r#"{"data": "Empty braces: \u{}"}"#,
];
for input in invalid_cases {
let result = jsonb_string_to_serde_value(input);
assert!(result.is_err());
}
Ok(())
}
#[test]
fn test_json_type_include() {
fn test(this: &JsonNativeType, that: &JsonNativeType, expected: bool) {
@@ -771,16 +650,15 @@ mod tests {
"list": [1, 2, 3],
"object": {"a": 1}
}"#;
let expected =
r#"Json<{"hello":"<String>","list":["<Number>"],"object":{"a":"<Number>"}}>"#;
let expected = r#"Json<Object{"hello": String, "list": Array[Number(I64)], "object": Object{"a": Number(I64)}}>"#;
test(json, json_type, Ok(expected))?;
// cannot merge with other non-object json values:
let jsons = [r#""s""#, "1", "[1]"];
let expects = [
r#"Failed to merge JSON datatype: datatypes have conflict, this: {"hello":"<String>","list":["<Number>"],"object":{"a":"<Number>"}}, that: "<String>""#,
r#"Failed to merge JSON datatype: datatypes have conflict, this: {"hello":"<String>","list":["<Number>"],"object":{"a":"<Number>"}}, that: "<Number>""#,
r#"Failed to merge JSON datatype: datatypes have conflict, this: {"hello":"<String>","list":["<Number>"],"object":{"a":"<Number>"}}, that: ["<Number>"]"#,
r#"Failed to merge JSON datatype: datatypes have conflict, this: Object{"hello": String, "list": Array[Number(I64)], "object": Object{"a": Number(I64)}}, that: String"#,
r#"Failed to merge JSON datatype: datatypes have conflict, this: Object{"hello": String, "list": Array[Number(I64)], "object": Object{"a": Number(I64)}}, that: Number(I64)"#,
r#"Failed to merge JSON datatype: datatypes have conflict, this: Object{"hello": String, "list": Array[Number(I64)], "object": Object{"a": Number(I64)}}, that: Array[Number(I64)]"#,
];
for (json, expect) in jsons.into_iter().zip(expects.into_iter()) {
test(json, json_type, Err(expect))?;
@@ -792,7 +670,7 @@ mod tests {
"float": 0.123,
"no": 42
}"#;
let expected = r#"Failed to merge JSON datatype: datatypes have conflict, this: "<String>", that: "<Number>""#;
let expected = r#"Failed to merge JSON datatype: datatypes have conflict, this: String, that: Number(I64)"#;
test(json, json_type, Err(expected))?;
// can merge with another json object:
@@ -801,7 +679,7 @@ mod tests {
"float": 0.123,
"int": 42
}"#;
let expected = r#"Json<{"float":"<Number>","hello":"<String>","int":"<Number>","list":["<Number>"],"object":{"a":"<Number>"}}>"#;
let expected = r#"Json<Object{"float": Number(F64), "hello": String, "int": Number(I64), "list": Array[Number(I64)], "object": Object{"a": Number(I64)}}>"#;
test(json, json_type, Ok(expected))?;
// can merge with some complex nested json object:
@@ -811,7 +689,7 @@ mod tests {
"float": 0.456,
"int": 0
}"#;
let expected = r#"Json<{"float":"<Number>","hello":"<String>","int":"<Number>","list":["<Number>"],"object":{"a":"<Number>","foo":"<String>","l":["<String>"],"o":{"key":"<String>"}}}>"#;
let expected = r#"Json<Object{"float": Number(F64), "hello": String, "int": Number(I64), "list": Array[Number(I64)], "object": Object{"a": Number(I64), "foo": String, "l": Array[String], "o": Object{"key": String}}}>"#;
test(json, json_type, Ok(expected))?;
Ok(())

View File

@@ -321,10 +321,10 @@ mod tests {
Ok(()),
Ok(()),
Err(
r#"Failed to merge JSON datatype: datatypes have conflict, this: "<Number>", that: "<String>""#,
"Failed to merge JSON datatype: datatypes have conflict, this: Number(I64), that: String",
),
Err(
r#"Failed to merge JSON datatype: datatypes have conflict, this: "<Number>", that: ["<Bool>"]"#,
"Failed to merge JSON datatype: datatypes have conflict, this: Number(I64), that: Array[Bool]",
),
];
let mut builder = JsonVectorBuilder::new(JsonNativeType::Null, 1);
@@ -396,12 +396,12 @@ mod tests {
// test children builders:
assert_eq!(builder.builders.len(), 6);
let expect_types = [
r#"Json<{"list":["<Number>"],"s":"<String>"}>"#,
r#"Json<{"float":"<Number>","s":"<String>"}>"#,
r#"Json<{"float":"<Number>","int":"<Number>"}>"#,
r#"Json<{"int":"<Number>","object":{"hello":"<String>","timestamp":"<Number>"}}>"#,
r#"Json<{"nested":{"a":{"b":{"b":{"a":"<String>"}}}},"object":{"timestamp":"<Number>"}}>"#,
r#"Json<{"nested":{"a":{"b":{"a":{"b":"<String>"}}}},"object":{"timestamp":"<Number>"}}>"#,
r#"Json<Object{"list": Array[Number(I64)], "s": String}>"#,
r#"Json<Object{"float": Number(F64), "s": String}>"#,
r#"Json<Object{"float": Number(F64), "int": Number(I64)}>"#,
r#"Json<Object{"int": Number(I64), "object": Object{"hello": String, "timestamp": Number(I64)}}>"#,
r#"Json<Object{"nested": Object{"a": Object{"b": Object{"b": Object{"a": String}}}}, "object": Object{"timestamp": Number(I64)}}>"#,
r#"Json<Object{"nested": Object{"a": Object{"b": Object{"a": Object{"b": String}}}}, "object": Object{"timestamp": Number(I64)}}>"#,
];
let expect_vectors = [
r#"
@@ -456,7 +456,7 @@ mod tests {
}
// test final merged json type:
let expected = r#"Json<{"float":"<Number>","int":"<Number>","list":["<Number>"],"nested":{"a":{"b":{"a":{"b":"<String>"},"b":{"a":"<String>"}}}},"object":{"hello":"<String>","timestamp":"<Number>"},"s":"<String>"}>"#;
let expected = r#"Json<Object{"float": Number(F64), "int": Number(I64), "list": Array[Number(I64)], "nested": Object{"a": Object{"b": Object{"a": Object{"b": String}, "b": Object{"a": String}}}}, "object": Object{"hello": String, "timestamp": Number(I64)}, "s": String}>"#;
assert_eq!(builder.data_type().to_string(), expected);
// test final produced vector:

View File

@@ -299,7 +299,7 @@ impl Default for MetasrvOptions {
#[allow(deprecated)]
server_addr: String::new(),
store_addrs: vec!["127.0.0.1:2379".to_string()],
backend_tls: Some(TlsOption::prefer()),
backend_tls: None,
selector: SelectorType::default(),
enable_region_failover: false,
heartbeat_interval: distributed_time_constants::BASE_HEARTBEAT_INTERVAL,

View File

@@ -13,41 +13,18 @@
// limitations under the License.
use std::any::Any;
use std::collections::HashMap;
use common_procedure::{Context as ProcedureContext, ProcedureWithId, Status};
use serde::{Deserialize, Serialize};
use store_api::storage::RegionId;
use crate::error::Result;
use crate::procedure::repartition::collect::{Collect, ProcedureMeta};
use crate::procedure::repartition::group::RepartitionGroupProcedure;
use crate::procedure::repartition::plan::RegionDescriptor;
use crate::procedure::repartition::{self, Context, State};
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct Dispatch;
#[allow(dead_code)]
fn build_region_mapping(
source_regions: &[RegionDescriptor],
target_regions: &[RegionDescriptor],
transition_map: &[Vec<usize>],
) -> HashMap<RegionId, Vec<RegionId>> {
transition_map
.iter()
.enumerate()
.map(|(source_idx, indices)| {
let source_region = source_regions[source_idx].region_id;
let target_regions = indices
.iter()
.map(|&target_idx| target_regions[target_idx].region_id)
.collect::<Vec<_>>();
(source_region, target_regions)
})
.collect::<HashMap<RegionId, _>>()
}
#[async_trait::async_trait]
#[typetag::serde]
impl State for Dispatch {
@@ -60,19 +37,11 @@ impl State for Dispatch {
let mut procedures = Vec::with_capacity(ctx.persistent_ctx.plans.len());
let mut procedure_metas = Vec::with_capacity(ctx.persistent_ctx.plans.len());
for (plan_index, plan) in ctx.persistent_ctx.plans.iter().enumerate() {
let region_mapping = build_region_mapping(
&plan.source_regions,
&plan.target_regions,
&plan.transition_map,
);
let persistent_ctx = repartition::group::PersistentContext::new(
plan.group_id,
table_id,
ctx.persistent_ctx.catalog_name.clone(),
ctx.persistent_ctx.schema_name.clone(),
plan.source_regions.clone(),
plan.target_regions.clone(),
region_mapping,
);
let group_procedure = RepartitionGroupProcedure::new(persistent_ctx, ctx);

View File

@@ -12,34 +12,27 @@
// See the License for the specific language governing permissions and
// limitations under the License.
pub(crate) mod apply_staging_manifest;
pub(crate) mod enter_staging_region;
pub(crate) mod remap_manifest;
pub(crate) mod repartition_end;
pub(crate) mod repartition_start;
pub(crate) mod update_metadata;
pub(crate) mod utils;
use std::any::Any;
use std::collections::HashMap;
use std::fmt::Debug;
use std::time::Duration;
use common_error::ext::BoxedError;
use common_meta::DatanodeId;
use common_meta::cache_invalidator::CacheInvalidatorRef;
use common_meta::instruction::CacheIdent;
use common_meta::key::datanode_table::{DatanodeTableKey, DatanodeTableValue, RegionInfo};
use common_meta::key::table_route::TableRouteValue;
use common_meta::key::{DeserializedValueWithBytes, TableMetadataManagerRef};
use common_meta::lock_key::{CatalogLock, RegionLock, SchemaLock};
use common_meta::peer::Peer;
use common_meta::rpc::router::RegionRoute;
use common_procedure::error::ToJsonSnafu;
use common_procedure::{
Context as ProcedureContext, Error as ProcedureError, LockKey, Procedure,
Result as ProcedureResult, Status, StringKey, UserMetadata,
Context as ProcedureContext, LockKey, Procedure, Result as ProcedureResult, Status,
UserMetadata,
};
use common_telemetry::error;
use serde::{Deserialize, Serialize};
use snafu::{OptionExt, ResultExt};
use store_api::storage::{RegionId, TableId};
@@ -78,12 +71,6 @@ impl RepartitionGroupProcedure {
}
}
#[derive(Debug, Serialize)]
pub struct RepartitionGroupData<'a> {
persistent_ctx: &'a PersistentContext,
state: &'a dyn State,
}
#[async_trait::async_trait]
impl Procedure for RepartitionGroupProcedure {
fn type_name(&self) -> &str {
@@ -91,48 +78,27 @@ impl Procedure for RepartitionGroupProcedure {
}
async fn execute(&mut self, _ctx: &ProcedureContext) -> ProcedureResult<Status> {
let state = &mut self.state;
todo!()
}
match state.next(&mut self.context, _ctx).await {
Ok((next, status)) => {
*state = next;
Ok(status)
}
Err(e) => {
if e.is_retryable() {
Err(ProcedureError::retry_later(e))
} else {
error!(
e;
"Repartition group procedure failed, group id: {}, table id: {}",
self.context.persistent_ctx.group_id,
self.context.persistent_ctx.table_id,
);
Err(ProcedureError::external(e))
}
}
}
async fn rollback(&mut self, _: &ProcedureContext) -> ProcedureResult<()> {
todo!()
}
fn rollback_supported(&self) -> bool {
false
true
}
fn dump(&self) -> ProcedureResult<String> {
let data = RepartitionGroupData {
persistent_ctx: &self.context.persistent_ctx,
state: self.state.as_ref(),
};
serde_json::to_string(&data).context(ToJsonSnafu)
todo!()
}
fn lock_key(&self) -> LockKey {
LockKey::new(self.context.persistent_ctx.lock_key())
todo!()
}
fn user_metadata(&self) -> Option<UserMetadata> {
// TODO(weny): support user metadata.
None
todo!()
}
}
@@ -157,8 +123,8 @@ pub struct GroupPrepareResult {
pub target_routes: Vec<RegionRoute>,
/// The primary source region id (first source region), used for retrieving region options.
pub central_region: RegionId,
/// The peer where the primary source region is located.
pub central_region_datanode: Peer,
/// The datanode id where the primary source region is located.
pub central_region_datanode_id: DatanodeId,
}
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
@@ -166,59 +132,30 @@ pub struct PersistentContext {
pub group_id: GroupId,
/// The table id of the repartition group.
pub table_id: TableId,
/// The catalog name of the repartition group.
pub catalog_name: String,
/// The schema name of the repartition group.
pub schema_name: String,
/// The source regions of the repartition group.
pub sources: Vec<RegionDescriptor>,
/// The target regions of the repartition group.
pub targets: Vec<RegionDescriptor>,
/// For each `source region`, the corresponding
/// `target regions` that overlap with it.
pub region_mapping: HashMap<RegionId, Vec<RegionId>>,
/// The result of group prepare.
/// The value will be set in [RepartitionStart](crate::procedure::repartition::group::repartition_start::RepartitionStart) state.
pub group_prepare_result: Option<GroupPrepareResult>,
/// The staging manifest paths of the repartition group.
/// The value will be set in [RemapManifest](crate::procedure::repartition::group::remap_manifest::RemapManifest) state.
pub staging_manifest_paths: HashMap<RegionId, String>,
}
impl PersistentContext {
pub fn new(
group_id: GroupId,
table_id: TableId,
catalog_name: String,
schema_name: String,
sources: Vec<RegionDescriptor>,
targets: Vec<RegionDescriptor>,
region_mapping: HashMap<RegionId, Vec<RegionId>>,
) -> Self {
Self {
group_id,
table_id,
catalog_name,
schema_name,
sources,
targets,
region_mapping,
group_prepare_result: None,
staging_manifest_paths: HashMap::new(),
}
}
pub fn lock_key(&self) -> Vec<StringKey> {
let mut lock_keys = Vec::with_capacity(2 + self.sources.len());
lock_keys.extend([
CatalogLock::Read(&self.catalog_name).into(),
SchemaLock::read(&self.catalog_name, &self.schema_name).into(),
]);
for source in &self.sources {
lock_keys.push(RegionLock::Write(source.region_id).into());
}
lock_keys
}
}
impl Context {
@@ -316,7 +253,7 @@ impl Context {
// Safety: prepare result is set in [RepartitionStart] state.
let prepare_result = self.persistent_ctx.group_prepare_result.as_ref().unwrap();
let central_region_datanode_table_value = self
.get_datanode_table_value(table_id, prepare_result.central_region_datanode.id)
.get_datanode_table_value(table_id, prepare_result.central_region_datanode_id)
.await?;
let RegionInfo {
region_options,

View File

@@ -1,333 +0,0 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use std::any::Any;
use std::collections::HashMap;
use std::time::{Duration, Instant};
use api::v1::meta::MailboxMessage;
use common_meta::instruction::{
ApplyStagingManifestReply, ApplyStagingManifestsReply, Instruction, InstructionReply,
};
use common_meta::peer::Peer;
use common_meta::rpc::router::RegionRoute;
use common_procedure::{Context as ProcedureContext, Status};
use common_telemetry::info;
use futures::future::join_all;
use serde::{Deserialize, Serialize};
use snafu::{OptionExt, ResultExt, ensure};
use store_api::storage::RegionId;
use crate::error::{self, Error, Result};
use crate::handler::HeartbeatMailbox;
use crate::procedure::repartition::group::update_metadata::UpdateMetadata;
use crate::procedure::repartition::group::utils::{
HandleMultipleResult, group_region_routes_by_peer, handle_multiple_results,
};
use crate::procedure::repartition::group::{Context, State};
use crate::procedure::repartition::plan::RegionDescriptor;
use crate::service::mailbox::{Channel, MailboxRef};
#[derive(Debug, Serialize, Deserialize)]
pub struct ApplyStagingManifest;
#[async_trait::async_trait]
#[typetag::serde]
impl State for ApplyStagingManifest {
async fn next(
&mut self,
ctx: &mut Context,
_procedure_ctx: &ProcedureContext,
) -> Result<(Box<dyn State>, Status)> {
self.apply_staging_manifests(ctx).await?;
Ok((
Box::new(UpdateMetadata::ApplyStaging),
Status::executing(true),
))
}
fn as_any(&self) -> &dyn Any {
self
}
}
impl ApplyStagingManifest {
fn build_apply_staging_manifest_instructions(
staging_manifest_paths: &HashMap<RegionId, String>,
target_routes: &[RegionRoute],
targets: &[RegionDescriptor],
central_region_id: RegionId,
) -> Result<HashMap<Peer, Vec<common_meta::instruction::ApplyStagingManifest>>> {
let target_partition_expr_by_region = targets
.iter()
.map(|target| {
Ok((
target.region_id,
target
.partition_expr
.as_json_str()
.context(error::SerializePartitionExprSnafu)?,
))
})
.collect::<Result<HashMap<_, _>>>()?;
// Safety: `leader_peer` is set for all region routes, checked in `repartition_start`.
let target_region_routes_by_peer = group_region_routes_by_peer(target_routes);
let mut instructions = HashMap::with_capacity(target_region_routes_by_peer.len());
for (peer, region_ids) in target_region_routes_by_peer {
let apply_staging_manifests = region_ids
.into_iter()
.map(|region_id| common_meta::instruction::ApplyStagingManifest {
region_id,
partition_expr: target_partition_expr_by_region[&region_id].clone(),
central_region_id,
manifest_path: staging_manifest_paths[&region_id].clone(),
})
.collect();
instructions.insert(peer.clone(), apply_staging_manifests);
}
Ok(instructions)
}
#[allow(dead_code)]
async fn apply_staging_manifests(&self, ctx: &mut Context) -> Result<()> {
let table_id = ctx.persistent_ctx.table_id;
let group_id = ctx.persistent_ctx.group_id;
let staging_manifest_paths = &ctx.persistent_ctx.staging_manifest_paths;
// Safety: the group prepare result is set in the RepartitionStart state.
let prepare_result = ctx.persistent_ctx.group_prepare_result.as_ref().unwrap();
let targets = &ctx.persistent_ctx.targets;
let target_routes = &prepare_result.target_routes;
let central_region_id = prepare_result.central_region;
let instructions = Self::build_apply_staging_manifest_instructions(
staging_manifest_paths,
target_routes,
targets,
central_region_id,
)?;
let operation_timeout =
ctx.next_operation_timeout()
.context(error::ExceededDeadlineSnafu {
operation: "Apply staging manifests",
})?;
let (peers, tasks): (Vec<_>, Vec<_>) = instructions
.iter()
.map(|(peer, apply_staging_manifests)| {
(
peer,
Self::apply_staging_manifest(
&ctx.mailbox,
&ctx.server_addr,
peer,
apply_staging_manifests,
operation_timeout,
),
)
})
.unzip();
info!(
"Sent apply staging manifests instructions to peers: {:?} for repartition table {}, group id {}",
peers, table_id, group_id
);
let format_err_msg = |idx: usize, error: &Error| {
let peer = peers[idx];
format!(
"Failed to apply staging manifests on datanode {:?}, error: {:?}",
peer, error
)
};
// Waits for all tasks to complete.
let results = join_all(tasks).await;
let result = handle_multiple_results(&results);
match result {
HandleMultipleResult::AllSuccessful => Ok(()),
HandleMultipleResult::AllRetryable(retryable_errors) => error::RetryLaterSnafu {
reason: format!(
"All retryable errors during applying staging manifests for repartition table {}, group id {}: {:?}",
table_id, group_id,
retryable_errors
.iter()
.map(|(idx, error)| format_err_msg(*idx, error))
.collect::<Vec<_>>()
.join(",")
),
}
.fail(),
HandleMultipleResult::AllNonRetryable(non_retryable_errors) => error::UnexpectedSnafu {
violated: format!(
"All non retryable errors during applying staging manifests for repartition table {}, group id {}: {:?}",
table_id, group_id,
non_retryable_errors
.iter()
.map(|(idx, error)| format_err_msg(*idx, error))
.collect::<Vec<_>>()
.join(",")
),
}
.fail(),
HandleMultipleResult::PartialRetryable {
retryable_errors,
non_retryable_errors,
} => error::UnexpectedSnafu {
violated: format!(
"Partial retryable errors during applying staging manifests for repartition table {}, group id {}: {:?}, non retryable errors: {:?}",
table_id, group_id,
retryable_errors
.iter()
.map(|(idx, error)| format_err_msg(*idx, error))
.collect::<Vec<_>>()
.join(","),
non_retryable_errors
.iter()
.map(|(idx, error)| format_err_msg(*idx, error))
.collect::<Vec<_>>()
.join(","),
),
}
.fail(),
}
}
async fn apply_staging_manifest(
mailbox: &MailboxRef,
server_addr: &str,
peer: &Peer,
apply_staging_manifests: &[common_meta::instruction::ApplyStagingManifest],
timeout: Duration,
) -> Result<()> {
let ch = Channel::Datanode(peer.id);
let instruction = Instruction::ApplyStagingManifests(apply_staging_manifests.to_vec());
let message = MailboxMessage::json_message(
&format!(
"Apply staging manifests for regions: {:?}",
apply_staging_manifests
.iter()
.map(|r| r.region_id)
.collect::<Vec<_>>()
),
&format!("Metasrv@{}", server_addr),
&format!("Datanode-{}@{}", peer.id, peer.addr),
common_time::util::current_time_millis(),
&instruction,
)
.with_context(|_| error::SerializeToJsonSnafu {
input: instruction.to_string(),
})?;
let now = Instant::now();
let receiver = mailbox.send(&ch, message, timeout).await;
let receiver = match receiver {
Ok(receiver) => receiver,
Err(error::Error::PusherNotFound { .. }) => error::RetryLaterSnafu {
reason: format!(
"Pusher not found for apply staging manifests on datanode {:?}, elapsed: {:?}",
peer,
now.elapsed()
),
}
.fail()?,
Err(err) => {
return Err(err);
}
};
match receiver.await {
Ok(msg) => {
let reply = HeartbeatMailbox::json_reply(&msg)?;
info!(
"Received apply staging manifests reply: {:?}, elapsed: {:?}",
reply,
now.elapsed()
);
let InstructionReply::ApplyStagingManifests(ApplyStagingManifestsReply { replies }) =
reply
else {
return error::UnexpectedInstructionReplySnafu {
mailbox_message: msg.to_string(),
reason: "expect apply staging manifests reply",
}
.fail();
};
for reply in replies {
Self::handle_apply_staging_manifest_reply(&reply, &now, peer)?;
}
Ok(())
}
Err(error::Error::MailboxTimeout { .. }) => {
let reason = format!(
"Mailbox received timeout for apply staging manifests on datanode {:?}, elapsed: {:?}",
peer,
now.elapsed()
);
error::RetryLaterSnafu { reason }.fail()
}
Err(err) => Err(err),
}
}
fn handle_apply_staging_manifest_reply(
ApplyStagingManifestReply {
region_id,
ready,
exists,
error,
}: &ApplyStagingManifestReply,
now: &Instant,
peer: &Peer,
) -> Result<()> {
ensure!(
exists,
error::UnexpectedSnafu {
violated: format!(
"Region {} doesn't exist on datanode {:?}, elapsed: {:?}",
region_id,
peer,
now.elapsed()
)
}
);
if error.is_some() {
return error::RetryLaterSnafu {
reason: format!(
"Failed to apply staging manifest on datanode {:?}, error: {:?}, elapsed: {:?}",
peer,
error,
now.elapsed()
),
}
.fail();
}
ensure!(
ready,
error::RetryLaterSnafu {
reason: format!(
"Region {} is still applying staging manifest on datanode {:?}, elapsed: {:?}",
region_id,
peer,
now.elapsed()
),
}
);
Ok(())
}
}

View File

@@ -29,7 +29,6 @@ use snafu::{OptionExt, ResultExt, ensure};
use crate::error::{self, Error, Result};
use crate::handler::HeartbeatMailbox;
use crate::procedure::repartition::group::remap_manifest::RemapManifest;
use crate::procedure::repartition::group::utils::{
HandleMultipleResult, group_region_routes_by_peer, handle_multiple_results,
};
@@ -50,7 +49,7 @@ impl State for EnterStagingRegion {
) -> Result<(Box<dyn State>, Status)> {
self.enter_staging_regions(ctx).await?;
Ok((Box::new(RemapManifest), Status::executing(true)))
Ok(Self::next_state())
}
fn as_any(&self) -> &dyn Any {
@@ -59,10 +58,16 @@ impl State for EnterStagingRegion {
}
impl EnterStagingRegion {
#[allow(dead_code)]
fn next_state() -> (Box<dyn State>, Status) {
// TODO(weny): change it later.
(Box::new(EnterStagingRegion), Status::executing(true))
}
fn build_enter_staging_instructions(
prepare_result: &GroupPrepareResult,
targets: &[RegionDescriptor],
) -> Result<HashMap<Peer, Vec<common_meta::instruction::EnterStagingRegion>>> {
) -> Result<HashMap<Peer, Instruction>> {
let target_partition_expr_by_region = targets
.iter()
.map(|target| {
@@ -88,7 +93,10 @@ impl EnterStagingRegion {
partition_expr: target_partition_expr_by_region[&region_id].clone(),
})
.collect();
instructions.insert(peer.clone(), enter_staging_regions);
instructions.insert(
peer.clone(),
Instruction::EnterStagingRegions(enter_staging_regions),
);
}
Ok(instructions)
@@ -109,14 +117,14 @@ impl EnterStagingRegion {
})?;
let (peers, tasks): (Vec<_>, Vec<_>) = instructions
.iter()
.map(|(peer, enter_staging_regions)| {
.map(|(peer, instruction)| {
(
peer,
Self::enter_staging_region(
&ctx.mailbox,
&ctx.server_addr,
peer,
enter_staging_regions,
instruction,
operation_timeout,
),
)
@@ -200,19 +208,12 @@ impl EnterStagingRegion {
mailbox: &MailboxRef,
server_addr: &str,
peer: &Peer,
enter_staging_regions: &[common_meta::instruction::EnterStagingRegion],
instruction: &Instruction,
timeout: Duration,
) -> Result<()> {
let ch = Channel::Datanode(peer.id);
let instruction = Instruction::EnterStagingRegions(enter_staging_regions.to_vec());
let message = MailboxMessage::json_message(
&format!(
"Enter staging regions: {:?}",
enter_staging_regions
.iter()
.map(|r| r.region_id)
.collect::<Vec<_>>()
),
&format!("Enter staging regions: {:?}", instruction),
&format!("Metasrv@{}", server_addr),
&format!("Datanode-{}@{}", peer.id, peer.addr),
common_time::util::current_time_millis(),
@@ -327,6 +328,7 @@ mod tests {
use std::assert_matches::assert_matches;
use std::time::Duration;
use common_meta::instruction::Instruction;
use common_meta::peer::Peer;
use common_meta::rpc::router::{Region, RegionRoute};
use store_api::storage::RegionId;
@@ -374,7 +376,7 @@ mod tests {
},
],
central_region: RegionId::new(table_id, 1),
central_region_datanode: Peer::empty(1),
central_region_datanode_id: 1,
};
let targets = test_targets();
let instructions =
@@ -382,7 +384,12 @@ mod tests {
.unwrap();
assert_eq!(instructions.len(), 2);
let instruction_1 = instructions.get(&Peer::empty(1)).unwrap().clone();
let instruction_1 = instructions
.get(&Peer::empty(1))
.unwrap()
.clone()
.into_enter_staging_regions()
.unwrap();
assert_eq!(
instruction_1,
vec![common_meta::instruction::EnterStagingRegion {
@@ -390,7 +397,12 @@ mod tests {
partition_expr: range_expr("x", 0, 10).as_json_str().unwrap(),
}]
);
let instruction_2 = instructions.get(&Peer::empty(2)).unwrap().clone();
let instruction_2 = instructions
.get(&Peer::empty(2))
.unwrap()
.clone()
.into_enter_staging_regions()
.unwrap();
assert_eq!(
instruction_2,
vec![common_meta::instruction::EnterStagingRegion {
@@ -405,17 +417,18 @@ mod tests {
let env = TestingEnv::new();
let server_addr = "localhost";
let peer = Peer::empty(1);
let enter_staging_regions = vec![common_meta::instruction::EnterStagingRegion {
region_id: RegionId::new(1024, 1),
partition_expr: range_expr("x", 0, 10).as_json_str().unwrap(),
}];
let instruction =
Instruction::EnterStagingRegions(vec![common_meta::instruction::EnterStagingRegion {
region_id: RegionId::new(1024, 1),
partition_expr: range_expr("x", 0, 10).as_json_str().unwrap(),
}]);
let timeout = Duration::from_secs(10);
let err = EnterStagingRegion::enter_staging_region(
env.mailbox_ctx.mailbox(),
server_addr,
&peer,
&enter_staging_regions,
&instruction,
timeout,
)
.await
@@ -434,10 +447,11 @@ mod tests {
.await;
let server_addr = "localhost";
let peer = Peer::empty(1);
let enter_staging_regions = vec![common_meta::instruction::EnterStagingRegion {
region_id: RegionId::new(1024, 1),
partition_expr: range_expr("x", 0, 10).as_json_str().unwrap(),
}];
let instruction =
Instruction::EnterStagingRegions(vec![common_meta::instruction::EnterStagingRegion {
region_id: RegionId::new(1024, 1),
partition_expr: range_expr("x", 0, 10).as_json_str().unwrap(),
}]);
let timeout = Duration::from_secs(10);
// Sends a timeout error.
@@ -449,7 +463,7 @@ mod tests {
env.mailbox_ctx.mailbox(),
server_addr,
&peer,
&enter_staging_regions,
&instruction,
timeout,
)
.await
@@ -465,10 +479,11 @@ mod tests {
let server_addr = "localhost";
let peer = Peer::empty(1);
let enter_staging_regions = vec![common_meta::instruction::EnterStagingRegion {
region_id: RegionId::new(1024, 1),
partition_expr: range_expr("x", 0, 10).as_json_str().unwrap(),
}];
let instruction =
Instruction::EnterStagingRegions(vec![common_meta::instruction::EnterStagingRegion {
region_id: RegionId::new(1024, 1),
partition_expr: range_expr("x", 0, 10).as_json_str().unwrap(),
}]);
let timeout = Duration::from_secs(10);
env.mailbox_ctx
@@ -483,7 +498,7 @@ mod tests {
env.mailbox_ctx.mailbox(),
server_addr,
&peer,
&enter_staging_regions,
&instruction,
timeout,
)
.await
@@ -501,10 +516,11 @@ mod tests {
.await;
let server_addr = "localhost";
let peer = Peer::empty(1);
let enter_staging_regions = vec![common_meta::instruction::EnterStagingRegion {
region_id: RegionId::new(1024, 1),
partition_expr: range_expr("x", 0, 10).as_json_str().unwrap(),
}];
let instruction =
Instruction::EnterStagingRegions(vec![common_meta::instruction::EnterStagingRegion {
region_id: RegionId::new(1024, 1),
partition_expr: range_expr("x", 0, 10).as_json_str().unwrap(),
}]);
let timeout = Duration::from_secs(10);
// Sends a failed reply.
@@ -522,7 +538,7 @@ mod tests {
env.mailbox_ctx.mailbox(),
server_addr,
&peer,
&enter_staging_regions,
&instruction,
timeout,
)
.await
@@ -549,7 +565,7 @@ mod tests {
env.mailbox_ctx.mailbox(),
server_addr,
&peer,
&enter_staging_regions,
&instruction,
timeout,
)
.await
@@ -580,7 +596,7 @@ mod tests {
},
],
central_region: RegionId::new(table_id, 1),
central_region_datanode: Peer::empty(1),
central_region_datanode_id: 1,
}
}

View File

@@ -1,222 +0,0 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use std::any::Any;
use std::collections::HashMap;
use std::time::{Duration, Instant};
use api::v1::meta::MailboxMessage;
use common_meta::instruction::{Instruction, InstructionReply, RemapManifestReply};
use common_meta::peer::Peer;
use common_procedure::{Context as ProcedureContext, Status};
use common_telemetry::{info, warn};
use serde::{Deserialize, Serialize};
use snafu::{OptionExt, ResultExt, ensure};
use store_api::storage::RegionId;
use crate::error::{self, Result};
use crate::handler::HeartbeatMailbox;
use crate::procedure::repartition::group::apply_staging_manifest::ApplyStagingManifest;
use crate::procedure::repartition::group::{Context, State};
use crate::procedure::repartition::plan::RegionDescriptor;
use crate::service::mailbox::{Channel, MailboxRef};
#[derive(Debug, Serialize, Deserialize)]
pub(crate) struct RemapManifest;
#[async_trait::async_trait]
#[typetag::serde]
impl State for RemapManifest {
async fn next(
&mut self,
ctx: &mut Context,
_procedure_ctx: &ProcedureContext,
) -> Result<(Box<dyn State>, Status)> {
let prepare_result = ctx.persistent_ctx.group_prepare_result.as_ref().unwrap();
let remap = Self::build_remap_manifest_instructions(
&ctx.persistent_ctx.sources,
&ctx.persistent_ctx.targets,
&ctx.persistent_ctx.region_mapping,
prepare_result.central_region,
)?;
let operation_timeout =
ctx.next_operation_timeout()
.context(error::ExceededDeadlineSnafu {
operation: "Remap manifests",
})?;
let manifest_paths = Self::remap_manifests(
&ctx.mailbox,
&ctx.server_addr,
&prepare_result.central_region_datanode,
&remap,
operation_timeout,
)
.await?;
let table_id = ctx.persistent_ctx.table_id;
let group_id = ctx.persistent_ctx.group_id;
if manifest_paths.len() != ctx.persistent_ctx.targets.len() {
warn!(
"Mismatch in manifest paths count: expected {}, got {}. This occurred during remapping manifests for group {} and table {}.",
ctx.persistent_ctx.targets.len(),
manifest_paths.len(),
group_id,
table_id
);
}
ctx.persistent_ctx.staging_manifest_paths = manifest_paths;
Ok((Box::new(ApplyStagingManifest), Status::executing(true)))
}
fn as_any(&self) -> &dyn Any {
self
}
}
impl RemapManifest {
fn build_remap_manifest_instructions(
source_regions: &[RegionDescriptor],
target_regions: &[RegionDescriptor],
region_mapping: &HashMap<RegionId, Vec<RegionId>>,
central_region_id: RegionId,
) -> Result<common_meta::instruction::RemapManifest> {
let new_partition_exprs = target_regions
.iter()
.map(|r| {
Ok((
r.region_id,
r.partition_expr
.as_json_str()
.context(error::SerializePartitionExprSnafu)?,
))
})
.collect::<Result<HashMap<RegionId, String>>>()?;
Ok(common_meta::instruction::RemapManifest {
region_id: central_region_id,
input_regions: source_regions.iter().map(|r| r.region_id).collect(),
region_mapping: region_mapping.clone(),
new_partition_exprs,
})
}
async fn remap_manifests(
mailbox: &MailboxRef,
server_addr: &str,
peer: &Peer,
remap: &common_meta::instruction::RemapManifest,
timeout: Duration,
) -> Result<HashMap<RegionId, String>> {
let ch = Channel::Datanode(peer.id);
let instruction = Instruction::RemapManifest(remap.clone());
let message = MailboxMessage::json_message(
&format!(
"Remap manifests, central region: {}, input regions: {:?}",
remap.region_id, remap.input_regions
),
&format!("Metasrv@{}", server_addr),
&format!("Datanode-{}@{}", peer.id, peer.addr),
common_time::util::current_time_millis(),
&instruction,
)
.with_context(|_| error::SerializeToJsonSnafu {
input: instruction.to_string(),
})?;
let now = Instant::now();
let receiver = mailbox.send(&ch, message, timeout).await;
let receiver = match receiver {
Ok(receiver) => receiver,
Err(error::Error::PusherNotFound { .. }) => error::RetryLaterSnafu {
reason: format!(
"Pusher not found for remap manifests on datanode {:?}, elapsed: {:?}",
peer,
now.elapsed()
),
}
.fail()?,
Err(err) => {
return Err(err);
}
};
match receiver.await {
Ok(msg) => {
let reply = HeartbeatMailbox::json_reply(&msg)?;
info!(
"Received remap manifest reply: {:?}, elapsed: {:?}",
reply,
now.elapsed()
);
let InstructionReply::RemapManifest(reply) = reply else {
return error::UnexpectedInstructionReplySnafu {
mailbox_message: msg.to_string(),
reason: "expect remap manifest reply",
}
.fail();
};
Self::handle_remap_manifest_reply(remap.region_id, reply, &now, peer)
}
Err(error::Error::MailboxTimeout { .. }) => {
let reason = format!(
"Mailbox received timeout for remap manifests on datanode {:?}, elapsed: {:?}",
peer,
now.elapsed()
);
error::RetryLaterSnafu { reason }.fail()
}
Err(err) => Err(err),
}
}
fn handle_remap_manifest_reply(
region_id: RegionId,
RemapManifestReply {
exists,
manifest_paths,
error,
}: RemapManifestReply,
now: &Instant,
peer: &Peer,
) -> Result<HashMap<RegionId, String>> {
ensure!(
exists,
error::UnexpectedSnafu {
violated: format!(
"Region {} doesn't exist on datanode {:?}, elapsed: {:?}",
region_id,
peer,
now.elapsed()
)
}
);
if error.is_some() {
return error::RetryLaterSnafu {
reason: format!(
"Failed to remap manifest on datanode {:?}, error: {:?}, elapsed: {:?}",
peer,
error,
now.elapsed()
),
}
.fail();
}
Ok(manifest_paths)
}
}

View File

@@ -1,40 +0,0 @@
// Copyright 2023 Greptime Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use std::any::Any;
use common_procedure::{Context as ProcedureContext, Status};
use serde::{Deserialize, Serialize};
use crate::error::Result;
use crate::procedure::repartition::group::{Context, State};
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct RepartitionEnd;
#[async_trait::async_trait]
#[typetag::serde]
impl State for RepartitionEnd {
async fn next(
&mut self,
_ctx: &mut Context,
_procedure_ctx: &ProcedureContext,
) -> Result<(Box<dyn State>, Status)> {
Ok((Box::new(RepartitionEnd), Status::done()))
}
fn as_any(&self) -> &dyn Any {
self
}
}

View File

@@ -22,7 +22,6 @@ use serde::{Deserialize, Serialize};
use snafu::{OptionExt, ResultExt, ensure};
use crate::error::{self, Result};
use crate::procedure::repartition::group::update_metadata::UpdateMetadata;
use crate::procedure::repartition::group::{
Context, GroupId, GroupPrepareResult, State, region_routes,
};
@@ -110,7 +109,7 @@ impl RepartitionStart {
);
}
let central_region = sources[0].region_id;
let central_region_datanode = source_region_routes[0]
let central_region_datanode_id = source_region_routes[0]
.leader_peer
.as_ref()
.context(error::UnexpectedSnafu {
@@ -119,22 +118,20 @@ impl RepartitionStart {
central_region
),
})?
.clone();
.id;
Ok(GroupPrepareResult {
source_routes: source_region_routes,
target_routes: target_region_routes,
central_region,
central_region_datanode,
central_region_datanode_id,
})
}
#[allow(dead_code)]
fn next_state() -> (Box<dyn State>, Status) {
(
Box::new(UpdateMetadata::ApplyStaging),
Status::executing(true),
)
// TODO(weny): change it later.
(Box::new(RepartitionStart), Status::executing(true))
}
}

View File

@@ -17,14 +17,12 @@ pub(crate) mod rollback_staging_region;
use std::any::Any;
use common_meta::lock_key::TableLock;
use common_procedure::{Context as ProcedureContext, Status};
use common_telemetry::warn;
use serde::{Deserialize, Serialize};
use crate::error::Result;
use crate::procedure::repartition::group::enter_staging_region::EnterStagingRegion;
use crate::procedure::repartition::group::repartition_end::RepartitionEnd;
use crate::procedure::repartition::group::repartition_start::RepartitionStart;
use crate::procedure::repartition::group::{Context, State};
#[derive(Debug, Serialize, Deserialize)]
@@ -35,16 +33,22 @@ pub enum UpdateMetadata {
RollbackStaging,
}
impl UpdateMetadata {
#[allow(dead_code)]
fn next_state() -> (Box<dyn State>, Status) {
// TODO(weny): change it later.
(Box::new(RepartitionStart), Status::executing(true))
}
}
#[async_trait::async_trait]
#[typetag::serde]
impl State for UpdateMetadata {
async fn next(
&mut self,
ctx: &mut Context,
procedure_ctx: &ProcedureContext,
_procedure_ctx: &ProcedureContext,
) -> Result<(Box<dyn State>, Status)> {
let table_lock = TableLock::Write(ctx.persistent_ctx.table_id).into();
let _guard = procedure_ctx.provider.acquire_lock(&table_lock).await;
match self {
UpdateMetadata::ApplyStaging => {
// TODO(weny): If all metadata have already been updated, skip applying staging regions.
@@ -55,7 +59,7 @@ impl State for UpdateMetadata {
"Failed to broadcast the invalidate table cache message during the apply staging regions, error: {err:?}"
);
};
Ok((Box::new(EnterStagingRegion), Status::executing(false)))
Ok(Self::next_state())
}
UpdateMetadata::RollbackStaging => {
self.rollback_staging_regions(ctx).await?;
@@ -65,7 +69,7 @@ impl State for UpdateMetadata {
"Failed to broadcast the invalidate table cache message during the rollback staging regions, error: {err:?}"
);
};
Ok((Box::new(RepartitionEnd), Status::executing(false)))
Ok(Self::next_state())
}
}
}

View File

@@ -41,9 +41,6 @@ pub struct AllocationPlanEntry {
pub regions_to_allocate: usize,
/// The number of regions that need to be deallocated (source count - target count, if positive).
pub regions_to_deallocate: usize,
/// For each `source_regions[k]`, the corresponding vector contains global
/// `target_partition_exprs` that overlap with it.
pub transition_map: Vec<Vec<usize>>,
}
/// A plan entry for the dispatch phase after region allocation,
@@ -60,9 +57,6 @@ pub struct RepartitionPlanEntry {
pub allocated_region_ids: Vec<RegionId>,
/// The region ids of the regions that are pending deallocation.
pub pending_deallocate_region_ids: Vec<RegionId>,
/// For each `source_regions[k]`, the corresponding vector contains global
/// `target_regions` that overlap with it.
pub transition_map: Vec<Vec<usize>>,
}
impl RepartitionPlanEntry {
@@ -77,7 +71,6 @@ impl RepartitionPlanEntry {
target_partition_exprs,
regions_to_allocate,
regions_to_deallocate,
transition_map,
}: &AllocationPlanEntry,
) -> Self {
debug_assert!(*regions_to_allocate == 0 && *regions_to_deallocate == 0);
@@ -96,7 +89,6 @@ impl RepartitionPlanEntry {
target_regions,
allocated_region_ids: vec![],
pending_deallocate_region_ids: vec![],
transition_map: transition_map.clone(),
}
}
}

View File

@@ -129,7 +129,6 @@ impl RepartitionStart {
target_partition_exprs,
regions_to_allocate,
regions_to_deallocate,
transition_map: subtask.transition_map,
}
})
.collect::<Vec<_>>()

View File

@@ -12,7 +12,6 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::collections::HashMap;
use std::sync::Arc;
use common_meta::key::{TableMetadataManager, TableMetadataManagerRef};
@@ -88,13 +87,9 @@ pub fn new_persistent_context(
) -> PersistentContext {
PersistentContext {
group_id: Uuid::new_v4(),
catalog_name: "test_catalog".to_string(),
schema_name: "test_schema".to_string(),
table_id,
sources,
targets,
region_mapping: HashMap::new(),
group_prepare_result: None,
staging_manifest_paths: HashMap::new(),
}
}

View File

@@ -13,7 +13,6 @@
// limitations under the License.
use common_meta::kv_backend::etcd::create_etcd_tls_options;
use common_telemetry::warn;
use etcd_client::{Client, ConnectOptions};
use servers::tls::{TlsMode, TlsOption};
use snafu::ResultExt;
@@ -39,12 +38,9 @@ pub async fn create_etcd_client_with_tls(
client_options.keep_alive_interval,
client_options.keep_alive_timeout,
);
let all_endpoints_use_https = etcd_endpoints.iter().all(|e| e.starts_with("https"));
if let Some(tls_config) = tls_config
&& let Some(tls_options) =
create_etcd_tls_options(&convert_tls_option(all_endpoints_use_https, tls_config))
.context(BuildTlsOptionsSnafu)?
&& let Some(tls_options) = create_etcd_tls_options(&convert_tls_option(tls_config))
.context(BuildTlsOptionsSnafu)?
{
connect_options = connect_options.with_tls(tls_options);
}
@@ -54,22 +50,9 @@ pub async fn create_etcd_client_with_tls(
.context(error::ConnectEtcdSnafu)
}
fn convert_tls_option(
all_endpoints_use_https: bool,
tls_option: &TlsOption,
) -> common_meta::kv_backend::etcd::TlsOption {
fn convert_tls_option(tls_option: &TlsOption) -> common_meta::kv_backend::etcd::TlsOption {
let mode = match tls_option.mode {
TlsMode::Disable => common_meta::kv_backend::etcd::TlsMode::Disable,
TlsMode::Prefer => {
if all_endpoints_use_https {
common_meta::kv_backend::etcd::TlsMode::Require
} else {
warn!(
"All endpoints use HTTP, TLS prefer mode is not supported, using disable mode"
);
common_meta::kv_backend::etcd::TlsMode::Disable
}
}
_ => common_meta::kv_backend::etcd::TlsMode::Require,
};
common_meta::kv_backend::etcd::TlsOption {

View File

@@ -13,7 +13,6 @@ workspace = true
ahash.workspace = true
api.workspace = true
arrow.workspace = true
arrow-schema.workspace = true
async-trait.workspace = true
catalog.workspace = true
chrono.workspace = true

View File

@@ -609,14 +609,10 @@ pub enum Error {
},
#[snafu(display(
"Multiple pipelines with different schemas found, but none under current schema. Please replicate one of them or delete until only one schema left. name: {}, current_schema: {}, schemas: {}",
name,
current_schema,
schemas,
"Multiple pipelines with different schemas found, but none under current schema. Please replicate one of them or delete until only one schema left. schemas: {}",
schemas
))]
MultiPipelineWithDiffSchema {
name: String,
current_schema: String,
schemas: String,
#[snafu(implicit)]
location: Location,
@@ -804,20 +800,6 @@ pub enum Error {
#[snafu(implicit)]
location: Location,
},
#[snafu(transparent)]
GreptimeProto {
source: api::error::Error,
#[snafu(implicit)]
location: Location,
},
#[snafu(transparent)]
Datatypes {
source: datatypes::error::Error,
#[snafu(implicit)]
location: Location,
},
}
pub type Result<T> = std::result::Result<T, Error>;
@@ -938,9 +920,6 @@ impl ErrorExt for Error {
FloatIsNan { .. }
| InvalidEpochForResolution { .. }
| UnsupportedTypeInPipeline { .. } => StatusCode::InvalidArguments,
GreptimeProto { source, .. } => source.status_code(),
Datatypes { source, .. } => source.status_code(),
}
}

View File

@@ -19,17 +19,13 @@ use std::collections::{BTreeMap, HashSet};
use std::sync::Arc;
use ahash::{HashMap, HashMapExt};
use api::helper::{ColumnDataTypeWrapper, encode_json_value};
use api::v1::column_def::{collect_column_options, options_from_column_schema};
use api::helper::proto_value_type;
use api::v1::column_data_type_extension::TypeExt;
use api::v1::value::ValueData;
use api::v1::{ColumnDataType, SemanticType};
use arrow_schema::extension::ExtensionType;
use api::v1::{ColumnDataType, ColumnDataTypeExtension, JsonTypeExtension, SemanticType};
use coerce::{coerce_columns, coerce_value};
use common_query::prelude::{greptime_timestamp, greptime_value};
use common_telemetry::warn;
use datatypes::data_type::ConcreteDataType;
use datatypes::extension::json::JsonExtensionType;
use datatypes::value::Value;
use greptime_proto::v1::{ColumnSchema, Row, Rows, Value as GreptimeValue};
use itertools::Itertools;
use jsonb::Number;
@@ -37,15 +33,12 @@ use once_cell::sync::OnceCell;
use serde_json as serde_json_crate;
use session::context::Channel;
use snafu::OptionExt;
use table::Table;
use vrl::prelude::{Bytes, VrlValueConvert};
use vrl::value::value::StdError;
use vrl::value::{KeyString, Value as VrlValue};
use crate::error::{
ArrayElementMustBeObjectSnafu, CoerceIncompatibleTypesSnafu,
IdentifyPipelineColumnTypeMismatchSnafu, InvalidTimestampSnafu, Result,
TimeIndexMustBeNonNullSnafu, TransformColumnNameMustBeUniqueSnafu,
ArrayElementMustBeObjectSnafu, IdentifyPipelineColumnTypeMismatchSnafu, InvalidTimestampSnafu,
Result, TimeIndexMustBeNonNullSnafu, TransformColumnNameMustBeUniqueSnafu,
TransformMultipleTimestampIndexSnafu, TransformTimestampIndexCountSnafu, ValueMustBeMapSnafu,
};
use crate::etl::PipelineDocVersion;
@@ -279,75 +272,15 @@ impl GreptimeTransformer {
}
}
#[derive(Clone)]
pub struct ColumnMetadata {
column_schema: datatypes::schema::ColumnSchema,
semantic_type: SemanticType,
}
impl From<ColumnSchema> for ColumnMetadata {
fn from(value: ColumnSchema) -> Self {
let datatype = value.datatype();
let semantic_type = value.semantic_type();
let ColumnSchema {
column_name,
datatype: _,
semantic_type: _,
datatype_extension,
options,
} = value;
let column_schema = datatypes::schema::ColumnSchema::new(
column_name,
ColumnDataTypeWrapper::new(datatype, datatype_extension).into(),
semantic_type != SemanticType::Timestamp,
);
let metadata = collect_column_options(options.as_ref());
let column_schema = column_schema.with_metadata(metadata);
Self {
column_schema,
semantic_type,
}
}
}
impl TryFrom<ColumnMetadata> for ColumnSchema {
type Error = api::error::Error;
fn try_from(value: ColumnMetadata) -> std::result::Result<Self, Self::Error> {
let ColumnMetadata {
column_schema,
semantic_type,
} = value;
let options = options_from_column_schema(&column_schema);
let (datatype, datatype_extension) =
ColumnDataTypeWrapper::try_from(column_schema.data_type).map(|x| x.into_parts())?;
Ok(ColumnSchema {
column_name: column_schema.name,
datatype: datatype as _,
semantic_type: semantic_type as _,
datatype_extension,
options,
})
}
}
/// This is used to record the current state schema information and a sequential cache of field names.
/// As you traverse the user input JSON, this will change.
/// It will record a superset of all user input schemas.
#[derive(Default)]
#[derive(Debug, Default)]
pub struct SchemaInfo {
/// schema info
pub schema: Vec<ColumnMetadata>,
pub schema: Vec<ColumnSchema>,
/// index of the column name
pub index: HashMap<String, usize>,
/// The pipeline's corresponding table (if already created). Useful to retrieve column schemas.
table: Option<Arc<Table>>,
}
impl SchemaInfo {
@@ -355,7 +288,6 @@ impl SchemaInfo {
Self {
schema: Vec::with_capacity(capacity),
index: HashMap::with_capacity(capacity),
table: None,
}
}
@@ -365,88 +297,46 @@ impl SchemaInfo {
index.insert(schema.column_name.clone(), i);
}
Self {
schema: schema_list.into_iter().map(Into::into).collect(),
schema: schema_list,
index,
table: None,
}
}
pub fn set_table(&mut self, table: Option<Arc<Table>>) {
self.table = table;
}
fn find_column_schema_in_table(&self, column_name: &str) -> Option<ColumnMetadata> {
if let Some(table) = &self.table
&& let Some(i) = table.schema_ref().column_index_by_name(column_name)
{
let column_schema = table.schema_ref().column_schemas()[i].clone();
let semantic_type = if column_schema.is_time_index() {
SemanticType::Timestamp
} else if table.table_info().meta.primary_key_indices.contains(&i) {
SemanticType::Tag
} else {
SemanticType::Field
};
Some(ColumnMetadata {
column_schema,
semantic_type,
})
} else {
None
}
}
pub fn column_schemas(&self) -> api::error::Result<Vec<ColumnSchema>> {
self.schema
.iter()
.map(|x| x.clone().try_into())
.collect::<api::error::Result<Vec<_>>>()
}
}
fn resolve_schema(
index: Option<usize>,
pipeline_context: &PipelineContext,
column: &str,
value_type: &ConcreteDataType,
value_data: ValueData,
column_schema: ColumnSchema,
row: &mut Vec<GreptimeValue>,
schema_info: &mut SchemaInfo,
) -> Result<()> {
if let Some(index) = index {
let column_type = &mut schema_info.schema[index].column_schema.data_type;
match (column_type, value_type) {
(column_type, value_type) if column_type == value_type => Ok(()),
(ConcreteDataType::Json(column_type), ConcreteDataType::Json(value_type))
if column_type.is_include(value_type) =>
{
Ok(())
let api_value = GreptimeValue {
value_data: Some(value_data),
};
// Safety unwrap is fine here because api_value is always valid
let value_column_data_type = proto_value_type(&api_value).unwrap();
// Safety unwrap is fine here because index is always valid
let schema_column_data_type = schema_info.schema.get(index).unwrap().datatype();
if value_column_data_type != schema_column_data_type {
IdentifyPipelineColumnTypeMismatchSnafu {
column: column_schema.column_name,
expected: schema_column_data_type.as_str_name(),
actual: value_column_data_type.as_str_name(),
}
(column_type, value_type) => IdentifyPipelineColumnTypeMismatchSnafu {
column,
expected: column_type.to_string(),
actual: value_type.to_string(),
}
.fail(),
.fail()
} else {
row[index] = api_value;
Ok(())
}
} else {
let column_schema = schema_info
.find_column_schema_in_table(column)
.unwrap_or_else(|| {
let semantic_type = decide_semantic(pipeline_context, column);
let column_schema = datatypes::schema::ColumnSchema::new(
column,
value_type.clone(),
semantic_type != SemanticType::Timestamp,
);
ColumnMetadata {
column_schema,
semantic_type,
}
});
let key = column.to_string();
let key = column_schema.column_name.clone();
schema_info.schema.push(column_schema);
schema_info.index.insert(key, schema_info.schema.len() - 1);
let api_value = GreptimeValue {
value_data: Some(value_data),
};
row.push(api_value);
Ok(())
}
}
@@ -591,11 +481,11 @@ pub(crate) fn values_to_row(
Ok(Row { values: row })
}
fn decide_semantic(p_ctx: &PipelineContext, column_name: &str) -> SemanticType {
fn decide_semantic(p_ctx: &PipelineContext, column_name: &str) -> i32 {
if p_ctx.channel == Channel::Prometheus && column_name != greptime_value() {
SemanticType::Tag
SemanticType::Tag as i32
} else {
SemanticType::Field
SemanticType::Field as i32
}
}
@@ -607,56 +497,55 @@ fn resolve_value(
p_ctx: &PipelineContext,
) -> Result<()> {
let index = schema_info.index.get(&column_name).copied();
let mut resolve_simple_type =
|value_data: ValueData, column_name: String, data_type: ColumnDataType| {
let semantic_type = decide_semantic(p_ctx, &column_name);
resolve_schema(
index,
value_data,
ColumnSchema {
column_name,
datatype: data_type as i32,
semantic_type,
datatype_extension: None,
options: None,
},
row,
schema_info,
)
};
let value_data = match value {
VrlValue::Null => None,
match value {
VrlValue::Null => {}
VrlValue::Integer(v) => {
// safe unwrap after type matched
resolve_schema(
index,
p_ctx,
&column_name,
&ConcreteDataType::int64_datatype(),
schema_info,
)?;
Some(ValueData::I64Value(v))
resolve_simple_type(ValueData::I64Value(v), column_name, ColumnDataType::Int64)?;
}
VrlValue::Float(v) => {
// safe unwrap after type matched
resolve_schema(
index,
p_ctx,
&column_name,
&ConcreteDataType::float64_datatype(),
schema_info,
resolve_simple_type(
ValueData::F64Value(v.into()),
column_name,
ColumnDataType::Float64,
)?;
Some(ValueData::F64Value(v.into()))
}
VrlValue::Boolean(v) => {
resolve_schema(
index,
p_ctx,
&column_name,
&ConcreteDataType::boolean_datatype(),
schema_info,
resolve_simple_type(
ValueData::BoolValue(v),
column_name,
ColumnDataType::Boolean,
)?;
Some(ValueData::BoolValue(v))
}
VrlValue::Bytes(v) => {
resolve_schema(
index,
p_ctx,
&column_name,
&ConcreteDataType::string_datatype(),
schema_info,
resolve_simple_type(
ValueData::StringValue(String::from_utf8_lossy_owned(v.to_vec())),
column_name,
ColumnDataType::String,
)?;
Some(ValueData::StringValue(String::from_utf8_lossy_owned(
v.to_vec(),
)))
}
VrlValue::Regex(v) => {
@@ -664,83 +553,42 @@ fn resolve_value(
"Persisting regex value in the table, this should not happen, column_name: {}",
column_name
);
resolve_schema(
index,
p_ctx,
&column_name,
&ConcreteDataType::string_datatype(),
schema_info,
resolve_simple_type(
ValueData::StringValue(v.to_string()),
column_name,
ColumnDataType::String,
)?;
Some(ValueData::StringValue(v.to_string()))
}
VrlValue::Timestamp(ts) => {
let ns = ts.timestamp_nanos_opt().context(InvalidTimestampSnafu {
input: ts.to_rfc3339(),
})?;
resolve_schema(
index,
p_ctx,
&column_name,
&ConcreteDataType::timestamp_nanosecond_datatype(),
schema_info,
resolve_simple_type(
ValueData::TimestampNanosecondValue(ns),
column_name,
ColumnDataType::TimestampNanosecond,
)?;
Some(ValueData::TimestampNanosecondValue(ns))
}
VrlValue::Array(_) | VrlValue::Object(_) => {
let is_json_native_type = schema_info
.find_column_schema_in_table(&column_name)
.is_some_and(|x| {
if let ConcreteDataType::Json(column_type) = &x.column_schema.data_type {
column_type.is_native_type()
} else {
false
}
});
let value = if is_json_native_type {
let json_extension_type: Option<JsonExtensionType> =
if let Some(x) = schema_info.find_column_schema_in_table(&column_name) {
x.column_schema.extension_type()?
} else {
None
};
let settings = json_extension_type
.and_then(|x| x.metadata().json_structure_settings.clone())
.unwrap_or_default();
let value: serde_json::Value = value.try_into().map_err(|e: StdError| {
CoerceIncompatibleTypesSnafu { msg: e.to_string() }.build()
})?;
let value = settings.encode(value)?;
resolve_schema(index, p_ctx, &column_name, &value.data_type(), schema_info)?;
let Value::Json(value) = value else {
unreachable!()
};
ValueData::JsonValue(encode_json_value(*value))
} else {
resolve_schema(
index,
p_ctx,
&column_name,
&ConcreteDataType::binary_datatype(),
schema_info,
)?;
let value = vrl_value_to_jsonb_value(&value);
ValueData::BinaryValue(value.to_vec())
};
Some(value)
let data = vrl_value_to_jsonb_value(&value);
resolve_schema(
index,
ValueData::BinaryValue(data.to_vec()),
ColumnSchema {
column_name,
datatype: ColumnDataType::Binary as i32,
semantic_type: SemanticType::Field as i32,
datatype_extension: Some(ColumnDataTypeExtension {
type_ext: Some(TypeExt::JsonType(JsonTypeExtension::JsonBinary.into())),
}),
options: None,
},
row,
schema_info,
)?;
}
};
let value = GreptimeValue { value_data };
if let Some(index) = index {
row[index] = value;
} else {
row.push(value);
}
Ok(())
}
@@ -778,24 +626,20 @@ fn identity_pipeline_inner(
let custom_ts = pipeline_ctx.pipeline_definition.get_custom_ts();
// set time index column schema first
let column_schema = datatypes::schema::ColumnSchema::new(
custom_ts
schema_info.schema.push(ColumnSchema {
column_name: custom_ts
.map(|ts| ts.get_column_name().to_string())
.unwrap_or_else(|| greptime_timestamp().to_string()),
custom_ts
.map(|c| ConcreteDataType::from(ColumnDataTypeWrapper::new(c.get_datatype(), None)))
.unwrap_or_else(|| {
if pipeline_ctx.channel == Channel::Prometheus {
ConcreteDataType::timestamp_millisecond_datatype()
} else {
ConcreteDataType::timestamp_nanosecond_datatype()
}
}),
false,
);
schema_info.schema.push(ColumnMetadata {
column_schema,
semantic_type: SemanticType::Timestamp,
datatype: custom_ts.map(|c| c.get_datatype()).unwrap_or_else(|| {
if pipeline_ctx.channel == Channel::Prometheus {
ColumnDataType::TimestampMillisecond
} else {
ColumnDataType::TimestampNanosecond
}
}) as i32,
semantic_type: SemanticType::Timestamp as i32,
datatype_extension: None,
options: None,
});
let mut opt_map = HashMap::new();
@@ -853,29 +697,28 @@ pub fn identity_pipeline(
input.push(result);
}
identity_pipeline_inner(input, pipeline_ctx).and_then(|(mut schema, opt_map)| {
identity_pipeline_inner(input, pipeline_ctx).map(|(mut schema, opt_map)| {
if let Some(table) = table {
let table_info = table.table_info();
for tag_name in table_info.meta.row_key_column_names() {
if let Some(index) = schema.index.get(tag_name) {
schema.schema[*index].semantic_type = SemanticType::Tag;
schema.schema[*index].semantic_type = SemanticType::Tag as i32;
}
}
}
let column_schemas = schema.column_schemas()?;
Ok(opt_map
opt_map
.into_iter()
.map(|(opt, rows)| {
(
opt,
Rows {
schema: column_schemas.clone(),
schema: schema.schema.clone(),
rows,
},
)
})
.collect::<HashMap<ContextOpt, Rows>>())
.collect::<HashMap<ContextOpt, Rows>>()
})
}
@@ -1007,7 +850,7 @@ mod tests {
assert!(rows.is_err());
assert_eq!(
rows.err().unwrap().to_string(),
"Column datatype mismatch. For column: score, expected datatype: Float64, actual datatype: String".to_string(),
"Column datatype mismatch. For column: score, expected datatype: FLOAT64, actual datatype: STRING".to_string(),
);
}
{
@@ -1036,7 +879,7 @@ mod tests {
assert!(rows.is_err());
assert_eq!(
rows.err().unwrap().to_string(),
"Column datatype mismatch. For column: score, expected datatype: Float64, actual datatype: Int64".to_string(),
"Column datatype mismatch. For column: score, expected datatype: FLOAT64, actual datatype: INT64".to_string(),
);
}
{
@@ -1099,7 +942,7 @@ mod tests {
.map(|(mut schema, mut rows)| {
for name in tag_column_names {
if let Some(index) = schema.index.get(&name) {
schema.schema[*index].semantic_type = SemanticType::Tag;
schema.schema[*index].semantic_type = SemanticType::Tag as i32;
}
}
@@ -1107,7 +950,7 @@ mod tests {
let rows = rows.remove(&ContextOpt::default()).unwrap();
Rows {
schema: schema.column_schemas().unwrap(),
schema: schema.schema,
rows,
}
});

View File

@@ -15,7 +15,6 @@
use std::sync::Arc;
use std::time::Duration;
use common_telemetry::debug;
use datatypes::timestamp::TimestampNanosecond;
use moka::sync::Cache;
@@ -34,18 +33,10 @@ const PIPELINES_CACHE_TTL: Duration = Duration::from_secs(10);
/// to encapsulate inner cache. Only public methods are exposed.
pub(crate) struct PipelineCache {
pipelines: Cache<String, Arc<Pipeline>>,
original_pipelines: Cache<String, PipelineContent>,
original_pipelines: Cache<String, (String, TimestampNanosecond)>,
/// If the pipeline table is invalid, we can use this cache to prevent failures when writing logs through the pipeline
/// The failover cache never expires, but it will be updated when the pipelines cache is updated.
failover_cache: Cache<String, PipelineContent>,
}
#[derive(Clone, Debug, PartialEq, Eq)]
pub struct PipelineContent {
pub name: String,
pub content: String,
pub version: TimestampNanosecond,
pub schema: String,
failover_cache: Cache<String, (String, TimestampNanosecond)>,
}
impl PipelineCache {
@@ -54,17 +45,12 @@ impl PipelineCache {
pipelines: Cache::builder()
.max_capacity(PIPELINES_CACHE_SIZE)
.time_to_live(PIPELINES_CACHE_TTL)
.name("pipelines")
.build(),
original_pipelines: Cache::builder()
.max_capacity(PIPELINES_CACHE_SIZE)
.time_to_live(PIPELINES_CACHE_TTL)
.name("original_pipelines")
.build(),
failover_cache: Cache::builder()
.max_capacity(PIPELINES_CACHE_SIZE)
.name("failover_cache")
.build(),
failover_cache: Cache::builder().max_capacity(PIPELINES_CACHE_SIZE).build(),
}
}
@@ -86,15 +72,19 @@ impl PipelineCache {
);
}
pub(crate) fn insert_pipeline_str_cache(&self, pipeline: &PipelineContent, with_latest: bool) {
let schema = pipeline.schema.as_str();
let name = pipeline.name.as_str();
let version = pipeline.version;
pub(crate) fn insert_pipeline_str_cache(
&self,
schema: &str,
name: &str,
version: PipelineVersion,
pipeline: (String, TimestampNanosecond),
with_latest: bool,
) {
insert_cache_generic(
&self.original_pipelines,
schema,
name,
Some(version),
version,
pipeline.clone(),
with_latest,
);
@@ -102,8 +92,8 @@ impl PipelineCache {
&self.failover_cache,
schema,
name,
Some(version),
pipeline.clone(),
version,
pipeline,
with_latest,
);
}
@@ -122,7 +112,7 @@ impl PipelineCache {
schema: &str,
name: &str,
version: PipelineVersion,
) -> Result<Option<PipelineContent>> {
) -> Result<Option<(String, TimestampNanosecond)>> {
get_cache_generic(&self.failover_cache, schema, name, version)
}
@@ -131,7 +121,7 @@ impl PipelineCache {
schema: &str,
name: &str,
version: PipelineVersion,
) -> Result<Option<PipelineContent>> {
) -> Result<Option<(String, TimestampNanosecond)>> {
get_cache_generic(&self.original_pipelines, schema, name, version)
}
@@ -184,13 +174,13 @@ fn get_cache_generic<T: Clone + Send + Sync + 'static>(
version: PipelineVersion,
) -> Result<Option<T>> {
// lets try empty schema first
let emp_key = generate_pipeline_cache_key(EMPTY_SCHEMA_NAME, name, version);
if let Some(value) = cache.get(&emp_key) {
let k = generate_pipeline_cache_key(EMPTY_SCHEMA_NAME, name, version);
if let Some(value) = cache.get(&k) {
return Ok(Some(value));
}
// use input schema
let schema_k = generate_pipeline_cache_key(schema, name, version);
if let Some(value) = cache.get(&schema_k) {
let k = generate_pipeline_cache_key(schema, name, version);
if let Some(value) = cache.get(&k) {
return Ok(Some(value));
}
@@ -203,28 +193,14 @@ fn get_cache_generic<T: Clone + Send + Sync + 'static>(
match ks.len() {
0 => Ok(None),
1 => {
let (_, value) = ks.remove(0);
Ok(Some(value))
}
_ => {
debug!(
"caches keys: {:?}, emp key: {:?}, schema key: {:?}, suffix key: {:?}",
cache.iter().map(|e| e.0).collect::<Vec<_>>(),
emp_key,
schema_k,
suffix_key
);
MultiPipelineWithDiffSchemaSnafu {
name: name.to_string(),
current_schema: schema.to_string(),
schemas: ks
.iter()
.filter_map(|(k, _)| k.split_once('/').map(|k| k.0))
.collect::<Vec<_>>()
.join(","),
}
.fail()?
1 => Ok(Some(ks.remove(0).1)),
_ => MultiPipelineWithDiffSchemaSnafu {
schemas: ks
.iter()
.filter_map(|(k, _)| k.split_once('/').map(|k| k.0))
.collect::<Vec<_>>()
.join(","),
}
.fail()?,
}
}

View File

@@ -220,7 +220,6 @@ impl PipelineOperator {
.observe(timer.elapsed().as_secs_f64())
})
.await
.map(|p| (p.content, p.version))
}
/// Insert a pipeline into the pipeline table.

View File

@@ -46,7 +46,7 @@ use crate::error::{
MultiPipelineWithDiffSchemaSnafu, PipelineNotFoundSnafu, RecordBatchLenNotMatchSnafu, Result,
};
use crate::etl::{Content, Pipeline, parse};
use crate::manager::pipeline_cache::{PipelineCache, PipelineContent};
use crate::manager::pipeline_cache::PipelineCache;
use crate::manager::{PipelineInfo, PipelineVersion};
use crate::metrics::METRIC_PIPELINE_TABLE_FIND_COUNT;
use crate::util::prepare_dataframe_conditions;
@@ -260,22 +260,17 @@ impl PipelineTable {
&self,
schema: &str,
name: &str,
input_version: PipelineVersion,
version: PipelineVersion,
) -> Result<Arc<Pipeline>> {
if let Some(pipeline) = self.cache.get_pipeline_cache(schema, name, input_version)? {
if let Some(pipeline) = self.cache.get_pipeline_cache(schema, name, version)? {
return Ok(pipeline);
}
let pipeline_content = self.get_pipeline_str(schema, name, input_version).await?;
let compiled_pipeline = Arc::new(Self::compile_pipeline(&pipeline_content.content)?);
let pipeline = self.get_pipeline_str(schema, name, version).await?;
let compiled_pipeline = Arc::new(Self::compile_pipeline(&pipeline.0)?);
self.cache.insert_pipeline_cache(
&pipeline_content.schema,
name,
Some(pipeline_content.version),
compiled_pipeline.clone(),
input_version.is_none(),
);
self.cache
.insert_pipeline_cache(schema, name, version, compiled_pipeline.clone(), false);
Ok(compiled_pipeline)
}
@@ -285,17 +280,14 @@ impl PipelineTable {
&self,
schema: &str,
name: &str,
input_version: PipelineVersion,
) -> Result<PipelineContent> {
if let Some(pipeline) = self
.cache
.get_pipeline_str_cache(schema, name, input_version)?
{
version: PipelineVersion,
) -> Result<(String, TimestampNanosecond)> {
if let Some(pipeline) = self.cache.get_pipeline_str_cache(schema, name, version)? {
return Ok(pipeline);
}
let mut pipeline_vec;
match self.find_pipeline(name, input_version).await {
match self.find_pipeline(name, version).await {
Ok(p) => {
METRIC_PIPELINE_TABLE_FIND_COUNT
.with_label_values(&["true"])
@@ -312,11 +304,8 @@ impl PipelineTable {
.inc();
return self
.cache
.get_failover_cache(schema, name, input_version)?
.context(PipelineNotFoundSnafu {
name,
version: input_version,
});
.get_failover_cache(schema, name, version)?
.ok_or(PipelineNotFoundSnafu { name, version }.build());
}
_ => {
// if other error, we should return it
@@ -327,40 +316,42 @@ impl PipelineTable {
};
ensure!(
!pipeline_vec.is_empty(),
PipelineNotFoundSnafu {
name,
version: input_version
}
PipelineNotFoundSnafu { name, version }
);
// if the result is exact one, use it
if pipeline_vec.len() == 1 {
let pipeline_content = pipeline_vec.remove(0);
self.cache
.insert_pipeline_str_cache(&pipeline_content, input_version.is_none());
return Ok(pipeline_content);
let (pipeline_content, found_schema, version) = pipeline_vec.remove(0);
let p = (pipeline_content, version);
self.cache.insert_pipeline_str_cache(
&found_schema,
name,
Some(version),
p.clone(),
false,
);
return Ok(p);
}
// check if there's empty schema pipeline
// if there isn't, check current schema
let pipeline = pipeline_vec
.iter()
.position(|v| v.schema == EMPTY_SCHEMA_NAME)
.or_else(|| pipeline_vec.iter().position(|v| v.schema == schema))
.map(|idx| pipeline_vec.remove(idx));
.find(|v| v.1 == EMPTY_SCHEMA_NAME)
.or_else(|| pipeline_vec.iter().find(|v| v.1 == schema));
// multiple pipeline with no empty or current schema
// throw an error
let pipeline_content = pipeline.with_context(|| MultiPipelineWithDiffSchemaSnafu {
name: name.to_string(),
current_schema: schema.to_string(),
schemas: pipeline_vec.iter().map(|v| v.schema.clone()).join(","),
})?;
let (pipeline_content, found_schema, version) =
pipeline.context(MultiPipelineWithDiffSchemaSnafu {
schemas: pipeline_vec.iter().map(|v| v.1.clone()).join(","),
})?;
let v = *version;
let p = (pipeline_content.clone(), v);
self.cache
.insert_pipeline_str_cache(&pipeline_content, input_version.is_none());
Ok(pipeline_content)
.insert_pipeline_str_cache(found_schema, name, Some(v), p.clone(), false);
Ok(p)
}
/// Insert a pipeline into the pipeline table and compile it.
@@ -387,15 +378,13 @@ impl PipelineTable {
true,
);
let pipeline_content = PipelineContent {
name: name.to_string(),
content: pipeline.to_string(),
version: TimestampNanosecond(version),
schema: EMPTY_SCHEMA_NAME.to_string(),
};
self.cache
.insert_pipeline_str_cache(&pipeline_content, true);
self.cache.insert_pipeline_str_cache(
EMPTY_SCHEMA_NAME,
name,
Some(TimestampNanosecond(version)),
(pipeline.to_owned(), TimestampNanosecond(version)),
true,
);
}
Ok((version, compiled_pipeline))
@@ -477,7 +466,7 @@ impl PipelineTable {
&self,
name: &str,
version: PipelineVersion,
) -> Result<Vec<PipelineContent>> {
) -> Result<Vec<(String, String, TimestampNanosecond)>> {
// 1. prepare dataframe
let dataframe = self
.query_engine
@@ -577,12 +566,11 @@ impl PipelineTable {
let len = pipeline_content.len();
for i in 0..len {
re.push(PipelineContent {
name: name.to_string(),
content: pipeline_content.value(i).to_string(),
version: TimestampNanosecond::new(pipeline_created_at.value(i)),
schema: pipeline_schema.value(i).to_string(),
});
re.push((
pipeline_content.value(i).to_string(),
pipeline_schema.value(i).to_string(),
TimestampNanosecond::new(pipeline_created_at.value(i)),
));
}
}

View File

@@ -61,7 +61,7 @@ pub fn parse_and_exec(input_str: &str, pipeline_yaml: &str) -> Rows {
}
Rows {
schema: schema_info.column_schemas().unwrap(),
schema: schema_info.schema.clone(),
rows,
}
}

View File

@@ -52,7 +52,7 @@ transform:
// check schema
assert_eq!(output.schema[0].column_name, "commit");
let type_id: i32 = ColumnDataType::Json.into();
let type_id: i32 = ColumnDataType::Binary.into();
assert_eq!(output.schema[0].datatype, type_id);
// check value
@@ -91,7 +91,7 @@ transform:
// check schema
assert_eq!(output.schema[0].column_name, "commit_json");
let type_id: i32 = ColumnDataType::Json.into();
let type_id: i32 = ColumnDataType::Binary.into();
assert_eq!(output.schema[0].datatype, type_id);
// check value
@@ -160,7 +160,7 @@ transform:
// check schema
assert_eq!(output.schema[0].column_name, "commit");
let type_id: i32 = ColumnDataType::Json.into();
let type_id: i32 = ColumnDataType::Binary.into();
assert_eq!(output.schema[0].datatype, type_id);
// check value

View File

@@ -281,18 +281,18 @@ struct PlanRewriter {
/// 2: Sort: t.pk1+t.pk2
/// 3. Projection: t.number, t.pk1, t.pk2
/// ```
/// `Sort` will make a column requirement for `t.pk1+t.pk2` at level 2.
/// `Sort` will make a column requirement for `t.pk1` at level 2.
/// Which making `Projection` at level 1 need to add a ref to `t.pk1` as well.
/// So that the expanded plan will be
/// ```ignore
/// Projection: t.number
/// MergeSort: t.pk1+t.pk2
/// MergeSort: t.pk1
/// MergeScan: remote_input=
/// Projection: t.number, "t.pk1+t.pk2" <--- the original `Projection` at level 1 get added with `t.pk1+t.pk2`
/// Sort: t.pk1+t.pk2
/// Projection: t.number, t.pk1, t.pk2
/// ```
/// Making `MergeSort` can have `t.pk1+t.pk2` as input.
/// Making `MergeSort` can have `t.pk1` as input.
/// Meanwhile `Projection` at level 3 doesn't need to add any new column because 3 > 2
/// and col requirements at level 2 is not applicable for level 3.
///
@@ -392,11 +392,10 @@ impl PlanRewriter {
&& ext_b.node.name() == MergeSortLogicalPlan::name()
{
// revert last `ConditionalCommutative` result for Sort plan in this case.
// also need to remove any column requirements made by the Sort Plan
// as it may refer to columns later no longer exist(rightfully) like by aggregate or projection
// `update_column_requirements` left unchanged because Sort won't generate
// new columns or remove existing columns.
self.stage.pop();
self.expand_on_next_part_cond_trans_commutative = false;
self.column_requirements.clear();
}
}
Commutativity::PartialCommutative => {
@@ -681,10 +680,6 @@ struct EnforceDistRequirementRewriter {
impl EnforceDistRequirementRewriter {
fn new(column_requirements: Vec<(HashSet<Column>, usize)>, cur_level: usize) -> Self {
debug!(
"Create EnforceDistRequirementRewriter with column_requirements: {:?} at cur_level: {}",
column_requirements, cur_level
);
Self {
column_requirements,
cur_level,
@@ -738,7 +733,7 @@ impl EnforceDistRequirementRewriter {
.filter(|a| !a.is_empty())
else {
return Err(datafusion_common::DataFusionError::Internal(format!(
"EnforceDistRequirementRewriter: no alias found for required column {original_col} at level {level} in current node's child plan: \n{child} from original plan: \n{original}",
"EnforceDistRequirementRewriter: no alias found for required column {original_col} in child plan {child} from original plan {original}",
)));
};

View File

@@ -777,67 +777,6 @@ fn expand_step_aggr_proj() {
assert_eq!(expected, result.to_string());
}
/// Make sure that `SeriesDivide` special handling correctly clean up column requirements from it's previous sort
#[test]
fn expand_complex_col_req_sort_pql() {
// use logging for better debugging
init_default_ut_logging();
let test_table = TestTable::table_with_name(0, "t".to_string());
let table_source = Arc::new(DefaultTableSource::new(Arc::new(
DfTableProviderAdapter::new(test_table),
)));
let plan = LogicalPlanBuilder::scan_with_filters("t", table_source.clone(), None, vec![])
.unwrap()
.sort(vec![
col("pk1").sort(true, false),
col("pk2").sort(true, false),
col("pk3").sort(true, false), // make some col req here
])
.unwrap()
.build()
.unwrap();
let plan = SeriesDivide::new(
vec!["pk1".to_string(), "pk2".to_string(), "pk3".to_string()],
"ts".to_string(),
plan,
);
let plan = LogicalPlan::Extension(datafusion_expr::Extension {
node: Arc::new(plan),
});
let plan = LogicalPlanBuilder::from(plan)
.aggregate(vec![col("pk1"), col("pk2")], vec![min(col("number"))])
.unwrap()
.sort(vec![
col("pk1").sort(true, false),
col("pk2").sort(true, false),
])
.unwrap()
.project(vec![col("pk1"), col("pk2")])
.unwrap()
.build()
.unwrap();
let config = ConfigOptions::default();
let result = DistPlannerAnalyzer {}.analyze(plan, &config).unwrap();
let expected = [
"Projection: t.pk1, t.pk2",
" MergeSort: t.pk1 ASC NULLS LAST, t.pk2 ASC NULLS LAST",
" MergeScan [is_placeholder=false, remote_input=[",
"Projection: t.pk1, t.pk2",
" Sort: t.pk1 ASC NULLS LAST, t.pk2 ASC NULLS LAST",
" Aggregate: groupBy=[[t.pk1, t.pk2]], aggr=[[min(t.number)]]",
r#" PromSeriesDivide: tags=["pk1", "pk2", "pk3"]"#,
" Sort: t.pk1 ASC NULLS LAST, t.pk2 ASC NULLS LAST, t.pk3 ASC NULLS LAST",
" TableScan: t",
"]]",
]
.join("\n");
assert_eq!(expected, result.to_string());
}
/// should only expand `Sort`, notice `Sort` before `Aggregate` usually can and
/// will be optimized out, and dist planner shouldn't handle that case, but
/// for now, still handle that be expanding the `Sort` node

View File

@@ -144,7 +144,7 @@ impl Categorizer {
}
}
// all group by expressions are partition columns can push down, unless
// another push down(including `Limit` or `Sort`) is already in progress(which will then prevent next cond commutative node from being push down).
// another push down(including `Limit` or `Sort`) is already in progress(which will then prvent next cond commutative node from being push down).
// TODO(discord9): This is a temporary solution(that works), a better description of
// commutativity is needed under this situation.
Commutativity::ConditionalCommutative(None)

View File

@@ -234,7 +234,7 @@ impl QueryEngineState {
rules.retain(|rule| rule.name() != name);
}
/// Optimize the logical plan by the extension analyzer rules.
/// Optimize the logical plan by the extension anayzer rules.
pub fn optimize_by_extension_rules(
&self,
plan: DfLogicalPlan,

View File

@@ -664,13 +664,6 @@ pub enum Error {
#[snafu(implicit)]
location: Location,
},
#[snafu(transparent)]
GreptimeProto {
source: api::error::Error,
#[snafu(implicit)]
location: Location,
},
}
pub type Result<T, E = Error> = std::result::Result<T, E>;
@@ -801,8 +794,6 @@ impl ErrorExt for Error {
Suspended { .. } => StatusCode::Suspended,
MemoryLimitExceeded { .. } => StatusCode::RateLimited,
GreptimeProto { source, .. } => source.status_code(),
}
}

View File

@@ -23,7 +23,6 @@ pub mod memory_limit;
pub mod prom_query_gateway;
pub mod region_server;
use std::any::Any;
use std::net::SocketAddr;
use std::time::Duration;
@@ -400,8 +399,4 @@ impl Server for GrpcServer {
fn bind_addr(&self) -> Option<SocketAddr> {
self.bind_addr
}
fn as_any(&self) -> &dyn Any {
self
}
}

View File

@@ -1285,10 +1285,6 @@ impl Server for HttpServer {
fn bind_addr(&self) -> Option<SocketAddr> {
self.bind_addr
}
fn as_any(&self) -> &dyn std::any::Any {
self
}
}
#[cfg(test)]

View File

@@ -31,7 +31,7 @@ use axum_extra::TypedHeader;
use common_catalog::consts::default_engine;
use common_error::ext::{BoxedError, ErrorExt};
use common_query::{Output, OutputData};
use common_telemetry::{error, warn};
use common_telemetry::{debug, error, warn};
use headers::ContentType;
use lazy_static::lazy_static;
use mime_guess::mime;
@@ -738,6 +738,11 @@ pub async fn log_ingester(
let value = extract_pipeline_value_by_content_type(content_type, payload, ignore_errors)?;
debug!(
"receiving logs: {:?}",
serde_json::to_string(&value).unwrap()
);
query_ctx.set_channel(Channel::Log);
let query_ctx = Arc::new(query_ctx);

View File

@@ -152,7 +152,7 @@ pub async fn loki_ingest(
rows.push(row);
}
let schemas = schema_info.column_schemas()?;
let schemas = schema_info.schema;
// fill Null for missing values
for row in rows.iter_mut() {
row.resize(schemas.len(), GreptimeValue::default());
@@ -746,16 +746,13 @@ fn process_labels(
} else {
// not exist
// add schema and append to values
schemas.push(
ColumnSchema {
column_name: k.clone(),
datatype: ColumnDataType::String.into(),
semantic_type: SemanticType::Tag.into(),
datatype_extension: None,
options: None,
}
.into(),
);
schemas.push(ColumnSchema {
column_name: k.clone(),
datatype: ColumnDataType::String.into(),
semantic_type: SemanticType::Tag.into(),
datatype_extension: None,
options: None,
});
column_indexer.insert(k, schemas.len() - 1);
row.push(GreptimeValue {

View File

@@ -12,7 +12,6 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::any::Any;
use std::future::Future;
use std::net::SocketAddr;
use std::sync::Arc;
@@ -266,8 +265,4 @@ impl Server for MysqlServer {
fn bind_addr(&self) -> Option<SocketAddr> {
self.bind_addr
}
fn as_any(&self) -> &dyn Any {
self
}
}

View File

@@ -381,7 +381,6 @@ fn extract_field_from_attr_and_combine_schema(
if let Some(index) = select_schema.index.get(key) {
let column_schema = &select_schema.schema[*index];
let column_schema: ColumnSchema = column_schema.clone().try_into()?;
// datatype of the same column name should be the same
ensure!(
column_schema.datatype == schema.datatype,
@@ -394,7 +393,7 @@ fn extract_field_from_attr_and_combine_schema(
);
extracted_values[*index] = value;
} else {
select_schema.schema.push(schema.into());
select_schema.schema.push(schema);
select_schema
.index
.insert(key.clone(), select_schema.schema.len() - 1);
@@ -481,7 +480,7 @@ fn parse_export_logs_service_request_to_rows(
let mut parse_ctx = ParseContext::new(select_info);
let mut rows = parse_resource(&mut parse_ctx, request.resource_logs)?;
schemas.extend(parse_ctx.select_schema.column_schemas()?);
schemas.extend(parse_ctx.select_schema.schema);
rows.iter_mut().for_each(|row| {
row.values.resize(schemas.len(), GreptimeValue::default());

View File

@@ -135,18 +135,12 @@ async fn run_custom_pipeline(
let mut schema_info = SchemaInfo::default();
schema_info
.schema
.push(time_index_column_schema(ts_name, timeunit).into());
.push(time_index_column_schema(ts_name, timeunit));
schema_info
}
};
let table = handler
.get_table(&table_name, query_ctx)
.await
.context(CatalogSnafu)?;
schema_info.set_table(table);
for pipeline_map in pipeline_maps {
let result = pipeline
.exec_mut(pipeline_map, pipeline_ctx, &mut schema_info)
@@ -200,7 +194,7 @@ async fn run_custom_pipeline(
RowInsertRequest {
rows: Some(Rows {
rows,
schema: schema_info.column_schemas()?,
schema: schema_info.schema.clone(),
}),
table_name: table_name.clone(),
},

View File

@@ -12,7 +12,6 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::any::Any;
use std::future::Future;
use std::net::SocketAddr;
use std::sync::Arc;
@@ -145,8 +144,4 @@ impl Server for PostgresServer {
fn bind_addr(&self) -> Option<SocketAddr> {
self.bind_addr
}
fn as_any(&self) -> &dyn Any {
self
}
}

View File

@@ -12,7 +12,6 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::any::Any;
use std::collections::HashMap;
use std::fmt::{Debug, Formatter};
use std::net::SocketAddr;
@@ -148,8 +147,6 @@ pub trait Server: Send + Sync {
fn bind_addr(&self) -> Option<SocketAddr> {
None
}
fn as_any(&self) -> &dyn Any;
}
struct AcceptTask {

View File

@@ -29,7 +29,7 @@ use strum::EnumString;
use crate::error::{InternalIoSnafu, Result};
/// TlsMode is used for Mysql and Postgres server start up.
#[derive(Debug, Default, Serialize, Deserialize, Clone, Copy, PartialEq, Eq, EnumString)]
#[derive(Debug, Default, Serialize, Deserialize, Clone, PartialEq, Eq, EnumString)]
#[serde(rename_all = "snake_case")]
pub enum TlsMode {
#[default]
@@ -91,17 +91,6 @@ impl TlsOption {
tls_option
}
/// Creates a new TLS option with the prefer mode.
pub fn prefer() -> Self {
Self {
mode: TlsMode::Prefer,
cert_path: String::new(),
key_path: String::new(),
ca_cert_path: String::new(),
watch: false,
}
}
/// Validates the TLS configuration.
///
/// Returns an error if:

View File

@@ -1,14 +1,14 @@
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+----------------------------+
| data | time_us |
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+----------------------------+
| {_raw: {"commit":{"cid":"bafyreidblutgvj75o4q4akzyyejedjj6l3it6hgqwee6jpwv2wqph5fsgm","record":{"$type":"app.bsky.feed.post","createdAt":"2024-11-21T16:09:27.095Z","langs":["en"],"reply":{"parent":{"cid":"bafyreibfglofvqou2yiqvwzk4rcgkhhxrbunyemshdjledgwymimqkg24e","uri":"at://did:plc:6tr6tuzlx2db3rduzr2d6r24/app.bsky.feed.post/3lbhqo2rtys2z"},"root":{"cid":"bafyreibfglofvqou2yiqvwzk4rcgkhhxrbunyemshdjledgwymimqkg24e","uri":"at://did:plc:6tr6tuzlx2db3rduzr2d6r24/app.bsky.feed.post/3lbhqo2rtys2z"}},"text":"aaaaah. LIght shines in a corner of WTF...."},"rev":"3lbhtytnn2k2f","rkey":"3lbhtyteurk2y"}}, commit.collection: app.bsky.feed.post, commit.operation: create, did: did:plc:yj3sjq3blzpynh27cumnp5ks, kind: commit, time_us: 1732206349000167} | 2024-11-21T16:25:49.000167 |
| {_raw: {"commit":{"cid":"bafyreia5l5vrkh5oj4cjyhcqby2dprhyvcyofo2q5562tijlae2pzih23m","record":{"$type":"app.bsky.feed.like","createdAt":"2024-11-21T16:25:46.221Z","subject":{"cid":"bafyreidjvrcmckkm765mct5fph36x7kupkfo35rjklbf2k76xkzwyiauge","uri":"at://did:plc:azrv4rcbws6kmcga4fsbphg2/app.bsky.feed.post/3lbgjdpbiec2l"}},"rev":"3lbhuvzds6d2a","rkey":"3lbhuvzdked2a"}}, commit.collection: app.bsky.feed.like, commit.operation: create, did: did:plc:3i4xf2v4wcnyktgv6satke64, kind: commit, time_us: 1732206349000644} | 2024-11-21T16:25:49.000644 |
| {_raw: {"commit":{"cid":"bafyreiew2p6cgirfaj45qoenm4fgumib7xoloclrap3jgkz5es7g7kby3i","record":{"$type":"app.bsky.graph.follow","createdAt":"2024-11-21T16:27:40.923Z","subject":"did:plc:r7cdh4sgzqbfdc6wcdxxti7c"},"rev":"3lbhuvze3gi2u","rkey":"3lbhuvzdtmi2u"}}, commit.collection: app.bsky.graph.follow, commit.operation: create, did: did:plc:gccfnqqizz4urhchsaie6jft, kind: commit, time_us: 1732206349001108} | 2024-11-21T16:25:49.001108 |
| {_raw: {"commit":{"cid":"bafyreidjix4dauj2afjlbzmhj3a7gwftcevvmmy6edww6vrjdbst26rkby","record":{"$type":"app.bsky.feed.like","createdAt":"2024-11-21T16:15:58.232Z","subject":{"cid":"bafyreiavpshyqzrlo5m7fqodjhs6jevweqnif4phasiwimv4a7mnsqi2fe","uri":"at://did:plc:fusulxqc52zbrc75fi6xrcof/app.bsky.feed.post/3lbhskq5zn22f"}},"rev":"3lbhueija5p22","rkey":"3lbhueiizcx22"}}, commit.collection: app.bsky.feed.like, commit.operation: create, did: did:plc:msxqf3twq7abtdw7dbfskphk, kind: commit, time_us: 1732206349001372} | 2024-11-21T16:25:49.001372 |
| {_raw: {"commit":{"cid":"bafyreigzdsdne3z2xxcakgisieyj7y47hj6eg7lj6v4q25ah5q2qotu5ku","record":{"$type":"app.bsky.feed.post","createdAt":"2024-11-21T16:09:27.254Z","langs":["en"],"reply":{"parent":{"cid":"bafyreih35fe2jj3gchmgk4amold4l6sfxd2sby5wrg3jrws5fkdypxrbg4","uri":"at://did:plc:6wx2gg5yqgvmlu35r6y3bk6d/app.bsky.feed.post/3lbhtj2eb4s2o"},"root":{"cid":"bafyreifipyt3vctd4ptuoicvio7rbr5xvjv4afwuggnd2prnmn55mu6luu","uri":"at://did:plc:474ldquxwzrlcvjhhbbk2wte/app.bsky.feed.post/3lbhdzrynik27"}},"text":"okay i take mine back because I hadnt heard this one yet^^"},"rev":"3lbhtytohxc2o","rkey":"3lbhtytjqzk2q"}}, commit.collection: app.bsky.feed.post, commit.operation: create, did: did:plc:l5o3qjrmfztir54cpwlv2eme, kind: commit, time_us: 1732206349001905} | 2024-11-21T16:25:49.001905 |
| {_raw: {"commit":{"cid":"bafyreiaa2vsdr4ckwjg4jq47zfd7mewidywfz3qh3dmglcd6ozi4xwdega","record":{"$type":"app.bsky.graph.follow","createdAt":"2024-11-21T16:15:21.495Z","subject":"did:plc:amsdn2tbjxo3xrwqneqhh4cm"},"rev":"3lbhudfo3yi2w","rkey":"3lbhudfnw4y2w"}}, commit.collection: app.bsky.graph.follow, commit.operation: create, did: did:plc:jkaaf5j2yb2pvpx3ualm3vbh, kind: commit, time_us: 1732206349002758} | 2024-11-21T16:25:49.002758 |
| {_raw: {"commit":{"cid":"bafyreihaatlpar3abtx6ck3kde2ksic6zzflk4ppduhf6dxurytqrv33ni","record":{"$type":"app.bsky.graph.follow","createdAt":"2024-11-21T16:18:39.913Z","subject":"did:plc:gf3vum7insztt5rxrpxdz2id"},"rev":"3lbhujcp4ix2n","rkey":"3lbhujcoxmp2n"}}, commit.collection: app.bsky.graph.follow, commit.operation: create, did: did:plc:tdwz2h4id5dxezvohftsmffu, kind: commit, time_us: 1732206349003106} | 2024-11-21T16:25:49.003106 |
| {_raw: {"commit":{"cid":"bafyreid5ycocp5zq2g7fcx2xxzxrbafuh7b5qhtwuwiomzo6vqila2cbpu","record":{"$type":"app.bsky.feed.repost","createdAt":"2024-11-21T16:23:36.714Z","subject":{"cid":"bafyreieaacfiobnuqvjhhsndyi5s3fd6krbzdduxsyrzfv43kczpcmkl6y","uri":"at://did:plc:o5q6dynpme4ndolc3heztasm/app.bsky.feed.post/3lbfli3qsoc2o"}},"rev":"3lbhus5vior2t","rkey":"3lbhus5vbtz2t"}}, commit.collection: app.bsky.feed.repost, commit.operation: create, did: did:plc:cdsd346mwow7aj3tgfkwsct3, kind: commit, time_us: 1732206349003461} | 2024-11-21T16:25:49.003461 |
| {_raw: {"commit":{"cid":"bafyreibugobcike72y4zxvdyz2oopyt6ywwqfielcwojkb27p7s6rlomgm","record":{"$type":"app.bsky.feed.post","createdAt":"2024-11-21T16:25:44.376Z","langs":["en"],"reply":{"parent":{"cid":"bafyreiaev27cfcxxvn2pdhrwwquzwgclujnulzbcfnn4p4fwgb6migjhw4","uri":"at://did:plc:zec6cslvgc3hhdatrhk6pq5p/app.bsky.feed.post/3lbhujvds4c2b"},"root":{"cid":"bafyreif7qjxhvecwnhlynijj6pf47jwvtkahsz3zh2kaipwu2bw2dxwaqq","uri":"at://did:plc:s4bwqchfzm6gjqfeb6mexgbu/app.bsky.feed.post/3lbhug53kkk2m"}},"text":"\n⌜ Blinking. She hadn't realized she spoke out loud. ⌟\n\n It was nothing like that — . I was only thinking . . . \n\n⌜ Trailing off, her mind occupied. ⌟\n"},"rev":"3lbhuvzeccx2w","rkey":"3lbhuvxf4qs2m"}}, commit.collection: app.bsky.feed.post, commit.operation: create, did: did:plc:s4bwqchfzm6gjqfeb6mexgbu, kind: commit, time_us: 1732206349003907} | 2024-11-21T16:25:49.003907 |
| {_raw: {"commit":{"cid":"bafyreidjk2svg2fdjiiwohmfmvp3hdxhpb33ycnixzbkyib5m6cocindxq","record":{"$type":"app.bsky.feed.like","createdAt":"2024-11-21T16:25:46.167Z","subject":{"cid":"bafyreiaumopip75nzx2xjbugtwemdppsyx54bd2odf6q45f3o7xkocgari","uri":"at://did:plc:ig2jv6gqup4t7gdq2pmanknw/app.bsky.feed.post/3lbhuvtlaec2c"}},"rev":"3lbhuvzedg52j","rkey":"3lbhuvzdyof2j"}}, commit.collection: app.bsky.feed.like, commit.operation: create, did: did:plc:hbc74dlsxhq53kp5oxges6d7, kind: commit, time_us: 1732206349004769} | 2024-11-21T16:25:49.004769 |
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+----------------------------+
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------------------+
| data | ts |
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------------------+
| {_raw: {"commit":{"cid":"bafyreidblutgvj75o4q4akzyyejedjj6l3it6hgqwee6jpwv2wqph5fsgm","record":{"$type":"app.bsky.feed.post","createdAt":"2024-11-21T16:09:27.095Z","langs":["en"],"reply":{"parent":{"cid":"bafyreibfglofvqou2yiqvwzk4rcgkhhxrbunyemshdjledgwymimqkg24e","uri":"at://did:plc:6tr6tuzlx2db3rduzr2d6r24/app.bsky.feed.post/3lbhqo2rtys2z"},"root":{"cid":"bafyreibfglofvqou2yiqvwzk4rcgkhhxrbunyemshdjledgwymimqkg24e","uri":"at://did:plc:6tr6tuzlx2db3rduzr2d6r24/app.bsky.feed.post/3lbhqo2rtys2z"}},"text":"aaaaah. LIght shines in a corner of WTF...."},"rev":"3lbhtytnn2k2f","rkey":"3lbhtyteurk2y"}}, commit.collection: app.bsky.feed.post, commit.operation: create, did: did:plc:yj3sjq3blzpynh27cumnp5ks, kind: commit, time_us: 1732206349000167} | 1970-01-01T00:00:00.001 |
| {_raw: {"commit":{"cid":"bafyreia5l5vrkh5oj4cjyhcqby2dprhyvcyofo2q5562tijlae2pzih23m","record":{"$type":"app.bsky.feed.like","createdAt":"2024-11-21T16:25:46.221Z","subject":{"cid":"bafyreidjvrcmckkm765mct5fph36x7kupkfo35rjklbf2k76xkzwyiauge","uri":"at://did:plc:azrv4rcbws6kmcga4fsbphg2/app.bsky.feed.post/3lbgjdpbiec2l"}},"rev":"3lbhuvzds6d2a","rkey":"3lbhuvzdked2a"}}, commit.collection: app.bsky.feed.like, commit.operation: create, did: did:plc:3i4xf2v4wcnyktgv6satke64, kind: commit, time_us: 1732206349000644} | 1970-01-01T00:00:00.002 |
| {_raw: {"commit":{"cid":"bafyreiew2p6cgirfaj45qoenm4fgumib7xoloclrap3jgkz5es7g7kby3i","record":{"$type":"app.bsky.graph.follow","createdAt":"2024-11-21T16:27:40.923Z","subject":"did:plc:r7cdh4sgzqbfdc6wcdxxti7c"},"rev":"3lbhuvze3gi2u","rkey":"3lbhuvzdtmi2u"}}, commit.collection: app.bsky.graph.follow, commit.operation: create, did: did:plc:gccfnqqizz4urhchsaie6jft, kind: commit, time_us: 1732206349001108} | 1970-01-01T00:00:00.003 |
| {_raw: {"commit":{"cid":"bafyreidjix4dauj2afjlbzmhj3a7gwftcevvmmy6edww6vrjdbst26rkby","record":{"$type":"app.bsky.feed.like","createdAt":"2024-11-21T16:15:58.232Z","subject":{"cid":"bafyreiavpshyqzrlo5m7fqodjhs6jevweqnif4phasiwimv4a7mnsqi2fe","uri":"at://did:plc:fusulxqc52zbrc75fi6xrcof/app.bsky.feed.post/3lbhskq5zn22f"}},"rev":"3lbhueija5p22","rkey":"3lbhueiizcx22"}}, commit.collection: app.bsky.feed.like, commit.operation: create, did: did:plc:msxqf3twq7abtdw7dbfskphk, kind: commit, time_us: 1732206349001372} | 1970-01-01T00:00:00.004 |
| {_raw: {"commit":{"cid":"bafyreigzdsdne3z2xxcakgisieyj7y47hj6eg7lj6v4q25ah5q2qotu5ku","record":{"$type":"app.bsky.feed.post","createdAt":"2024-11-21T16:09:27.254Z","langs":["en"],"reply":{"parent":{"cid":"bafyreih35fe2jj3gchmgk4amold4l6sfxd2sby5wrg3jrws5fkdypxrbg4","uri":"at://did:plc:6wx2gg5yqgvmlu35r6y3bk6d/app.bsky.feed.post/3lbhtj2eb4s2o"},"root":{"cid":"bafyreifipyt3vctd4ptuoicvio7rbr5xvjv4afwuggnd2prnmn55mu6luu","uri":"at://did:plc:474ldquxwzrlcvjhhbbk2wte/app.bsky.feed.post/3lbhdzrynik27"}},"text":"okay i take mine back because I hadnt heard this one yet^^"},"rev":"3lbhtytohxc2o","rkey":"3lbhtytjqzk2q"}}, commit.collection: app.bsky.feed.post, commit.operation: create, did: did:plc:l5o3qjrmfztir54cpwlv2eme, kind: commit, time_us: 1732206349001905} | 1970-01-01T00:00:00.005 |
| {_raw: {"commit":{"cid":"bafyreiaa2vsdr4ckwjg4jq47zfd7mewidywfz3qh3dmglcd6ozi4xwdega","record":{"$type":"app.bsky.graph.follow","createdAt":"2024-11-21T16:15:21.495Z","subject":"did:plc:amsdn2tbjxo3xrwqneqhh4cm"},"rev":"3lbhudfo3yi2w","rkey":"3lbhudfnw4y2w"}}, commit.collection: app.bsky.graph.follow, commit.operation: create, did: did:plc:jkaaf5j2yb2pvpx3ualm3vbh, kind: commit, time_us: 1732206349002758} | 1970-01-01T00:00:00.006 |
| {_raw: {"commit":{"cid":"bafyreihaatlpar3abtx6ck3kde2ksic6zzflk4ppduhf6dxurytqrv33ni","record":{"$type":"app.bsky.graph.follow","createdAt":"2024-11-21T16:18:39.913Z","subject":"did:plc:gf3vum7insztt5rxrpxdz2id"},"rev":"3lbhujcp4ix2n","rkey":"3lbhujcoxmp2n"}}, commit.collection: app.bsky.graph.follow, commit.operation: create, did: did:plc:tdwz2h4id5dxezvohftsmffu, kind: commit, time_us: 1732206349003106} | 1970-01-01T00:00:00.007 |
| {_raw: {"commit":{"cid":"bafyreid5ycocp5zq2g7fcx2xxzxrbafuh7b5qhtwuwiomzo6vqila2cbpu","record":{"$type":"app.bsky.feed.repost","createdAt":"2024-11-21T16:23:36.714Z","subject":{"cid":"bafyreieaacfiobnuqvjhhsndyi5s3fd6krbzdduxsyrzfv43kczpcmkl6y","uri":"at://did:plc:o5q6dynpme4ndolc3heztasm/app.bsky.feed.post/3lbfli3qsoc2o"}},"rev":"3lbhus5vior2t","rkey":"3lbhus5vbtz2t"}}, commit.collection: app.bsky.feed.repost, commit.operation: create, did: did:plc:cdsd346mwow7aj3tgfkwsct3, kind: commit, time_us: 1732206349003461} | 1970-01-01T00:00:00.008 |
| {_raw: {"commit":{"cid":"bafyreibugobcike72y4zxvdyz2oopyt6ywwqfielcwojkb27p7s6rlomgm","record":{"$type":"app.bsky.feed.post","createdAt":"2024-11-21T16:25:44.376Z","langs":["en"],"reply":{"parent":{"cid":"bafyreiaev27cfcxxvn2pdhrwwquzwgclujnulzbcfnn4p4fwgb6migjhw4","uri":"at://did:plc:zec6cslvgc3hhdatrhk6pq5p/app.bsky.feed.post/3lbhujvds4c2b"},"root":{"cid":"bafyreif7qjxhvecwnhlynijj6pf47jwvtkahsz3zh2kaipwu2bw2dxwaqq","uri":"at://did:plc:s4bwqchfzm6gjqfeb6mexgbu/app.bsky.feed.post/3lbhug53kkk2m"}},"text":"\n⌜ Blinking. She hadn't realized she spoke out loud. ⌟\n\n It was nothing like that — . I was only thinking . . . \n\n⌜ Trailing off, her mind occupied. ⌟\n"},"rev":"3lbhuvzeccx2w","rkey":"3lbhuvxf4qs2m"}}, commit.collection: app.bsky.feed.post, commit.operation: create, did: did:plc:s4bwqchfzm6gjqfeb6mexgbu, kind: commit, time_us: 1732206349003907} | 1970-01-01T00:00:00.009 |
| {_raw: {"commit":{"cid":"bafyreidjk2svg2fdjiiwohmfmvp3hdxhpb33ycnixzbkyib5m6cocindxq","record":{"$type":"app.bsky.feed.like","createdAt":"2024-11-21T16:25:46.167Z","subject":{"cid":"bafyreiaumopip75nzx2xjbugtwemdppsyx54bd2odf6q45f3o7xkocgari","uri":"at://did:plc:ig2jv6gqup4t7gdq2pmanknw/app.bsky.feed.post/3lbhuvtlaec2c"}},"rev":"3lbhuvzedg52j","rkey":"3lbhuvzdyof2j"}}, commit.collection: app.bsky.feed.like, commit.operation: create, did: did:plc:hbc74dlsxhq53kp5oxges6d7, kind: commit, time_us: 1732206349004769} | 1970-01-01T00:00:00.010 |
+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------------------+

View File

@@ -48,9 +48,9 @@ use flow::{FlownodeBuilder, FrontendClient, GrpcQueryHandlerWithBoxedError};
use frontend::frontend::Frontend;
use frontend::instance::builder::FrontendBuilder;
use frontend::instance::{Instance, StandaloneDatanodeManager};
use frontend::server::Services;
use meta_srv::metasrv::{FLOW_ID_SEQ, TABLE_ID_SEQ};
use servers::grpc::GrpcOptions;
use servers::server::ServerHandlers;
use snafu::ResultExt;
use standalone::options::StandaloneOptions;
@@ -249,7 +249,7 @@ impl GreptimeDbStandaloneBuilder {
procedure_executor.clone(),
Arc::new(ProcessManager::new(server_addr, None)),
)
.with_plugin(plugins.clone())
.with_plugin(plugins)
.try_build()
.await
.unwrap();
@@ -282,15 +282,14 @@ impl GreptimeDbStandaloneBuilder {
test_util::prepare_another_catalog_and_schema(&instance).await;
let servers = Services::new(opts.clone(), instance.clone(), plugins)
.build()
.unwrap();
let frontend = Frontend {
let mut frontend = Frontend {
instance,
servers,
servers: ServerHandlers::default(),
heartbeat_task: None,
};
frontend.start().await.unwrap();
GreptimeDbStandalone {
frontend: Arc::new(frontend),
opts,

View File

@@ -18,114 +18,14 @@ use std::{fs, io};
use common_test_util::find_workspace_path;
use frontend::instance::Instance;
use http::StatusCode;
use servers::http::test_helpers::TestClient;
use servers::http::{HTTP_SERVER, HttpServer};
use servers::server::ServerHandlers;
use tests_integration::standalone::GreptimeDbStandaloneBuilder;
use tests_integration::test_util::execute_sql_and_expect;
#[tokio::test(flavor = "multi_thread", worker_threads = 4)]
async fn test_load_jsonbench_data_by_pipeline() -> io::Result<()> {
#[tokio::test]
async fn test_load_jsonbench_data() {
common_telemetry::init_default_ut_logging();
let instance = GreptimeDbStandaloneBuilder::new("test_load_jsonbench_data_by_pipeline")
.build()
.await;
let frontend = instance.fe_instance();
let ServerHandlers::Init(handlers) = instance.frontend.server_handlers() else {
unreachable!()
};
let router = {
let handlers = handlers.lock().unwrap();
let server = handlers
.get(HTTP_SERVER)
.and_then(|x| x.0.as_any().downcast_ref::<HttpServer>())
.unwrap();
server.build(server.make_app()).unwrap()
};
let client = TestClient::new(router).await;
create_table(frontend).await;
desc_table(frontend).await;
create_pipeline(&client).await;
insert_data_by_pipeline(&client).await?;
query_data(frontend).await
}
async fn insert_data_by_pipeline(client: &TestClient) -> io::Result<()> {
let file = fs::read(find_workspace_path(
"tests-integration/resources/jsonbench-head-10.ndjson",
))?;
let response = client
.post("/v1/ingest?table=bluesky&pipeline_name=jsonbench")
.header("Content-Type", "text/plain")
.body(file)
.send()
.await;
assert_eq!(response.status(), StatusCode::OK);
let response = response.text().await;
// Note that this pattern also matches the inserted rows: "10".
let pattern = r#"{"output":[{"affectedrows":10}]"#;
assert!(response.starts_with(pattern));
Ok(())
}
async fn create_pipeline(client: &TestClient) {
let pipeline = r#"
version: 2
processors:
- json_parse:
fields:
- message, data
ignore_missing: true
- simple_extract:
fields:
- data, time_us
key: "time_us"
ignore_missing: false
- epoch:
fields:
- time_us
resolution: microsecond
- select:
fields:
- time_us
- data
transform:
- fields:
- time_us
type: epoch, us
index: timestamp
"#;
let response = client
.post("/v1/pipelines/jsonbench")
.header("Content-Type", "application/x-yaml")
.body(pipeline)
.send()
.await;
assert_eq!(response.status(), StatusCode::OK);
let response = response.text().await;
let pattern = r#"{"pipelines":[{"name":"jsonbench""#;
assert!(response.starts_with(pattern));
}
#[tokio::test(flavor = "multi_thread", worker_threads = 4)]
async fn test_load_jsonbench_data_by_sql() -> io::Result<()> {
common_telemetry::init_default_ut_logging();
let instance = GreptimeDbStandaloneBuilder::new("test_load_jsonbench_data_by_sql")
let instance = GreptimeDbStandaloneBuilder::new("test_load_jsonbench_data")
.build()
.await;
let frontend = instance.fe_instance();
@@ -134,9 +34,9 @@ async fn test_load_jsonbench_data_by_sql() -> io::Result<()> {
desc_table(frontend).await;
insert_data_by_sql(frontend).await?;
insert_data(frontend).await.unwrap();
query_data(frontend).await
query_data(frontend).await.unwrap();
}
async fn query_data(frontend: &Arc<Instance>) -> io::Result<()> {
@@ -146,21 +46,22 @@ async fn query_data(frontend: &Arc<Instance>) -> io::Result<()> {
| count(*) |
+----------+
| 10 |
+----------+"#;
+----------+
"#;
execute_sql_and_expect(frontend, sql, expected).await;
let sql = "SELECT * FROM bluesky ORDER BY time_us";
let sql = "SELECT * FROM bluesky ORDER BY ts";
let expected = fs::read_to_string(find_workspace_path(
"tests-integration/resources/jsonbench-select-all.txt",
))?;
execute_sql_and_expect(frontend, sql, &expected).await;
// query 1:
let sql = "
SELECT
json_get_string(data, '$.commit.collection') AS event, count() AS count
FROM bluesky
GROUP BY event
let sql = "\
SELECT \
json_get_string(data, '$.commit.collection') AS event, count() AS count \
FROM bluesky \
GROUP BY event \
ORDER BY count DESC, event ASC";
let expected = r#"
+-----------------------+-------+
@@ -174,16 +75,16 @@ ORDER BY count DESC, event ASC";
execute_sql_and_expect(frontend, sql, expected).await;
// query 2:
let sql = "
SELECT
json_get_string(data, '$.commit.collection') AS event,
count() AS count,
count(DISTINCT json_get_string(data, '$.did')) AS users
FROM bluesky
WHERE
(json_get_string(data, '$.kind') = 'commit') AND
(json_get_string(data, '$.commit.operation') = 'create')
GROUP BY event
let sql = "\
SELECT \
json_get_string(data, '$.commit.collection') AS event, \
count() AS count, \
count(DISTINCT json_get_string(data, '$.did')) AS users \
FROM bluesky \
WHERE \
(json_get_string(data, '$.kind') = 'commit') AND \
(json_get_string(data, '$.commit.operation') = 'create') \
GROUP BY event \
ORDER BY count DESC, event ASC";
let expected = r#"
+-----------------------+-------+-------+
@@ -197,18 +98,18 @@ ORDER BY count DESC, event ASC";
execute_sql_and_expect(frontend, sql, expected).await;
// query 3:
let sql = "
SELECT
json_get_string(data, '$.commit.collection') AS event,
date_part('hour', to_timestamp_micros(json_get_int(data, '$.time_us'))) as hour_of_day,
count() AS count
FROM bluesky
WHERE
(json_get_string(data, '$.kind') = 'commit') AND
(json_get_string(data, '$.commit.operation') = 'create') AND
json_get_string(data, '$.commit.collection') IN
('app.bsky.feed.post', 'app.bsky.feed.repost', 'app.bsky.feed.like')
GROUP BY event, hour_of_day
let sql = "\
SELECT \
json_get_string(data, '$.commit.collection') AS event, \
date_part('hour', to_timestamp_micros(json_get_int(data, '$.time_us'))) as hour_of_day, \
count() AS count \
FROM bluesky \
WHERE \
(json_get_string(data, '$.kind') = 'commit') AND \
(json_get_string(data, '$.commit.operation') = 'create') AND \
json_get_string(data, '$.commit.collection') IN \
('app.bsky.feed.post', 'app.bsky.feed.repost', 'app.bsky.feed.like') \
GROUP BY event, hour_of_day \
ORDER BY hour_of_day, event";
let expected = r#"
+----------------------+-------------+-------+
@@ -221,7 +122,7 @@ ORDER BY hour_of_day, event";
execute_sql_and_expect(frontend, sql, expected).await;
// query 4:
let sql = "
let sql = "\
SELECT
json_get_string(data, '$.did') as user_id,
min(to_timestamp_micros(json_get_int(data, '$.time_us'))) AS first_post_ts
@@ -273,23 +174,19 @@ LIMIT 3";
Ok(())
}
async fn insert_data_by_sql(frontend: &Arc<Instance>) -> io::Result<()> {
async fn insert_data(frontend: &Arc<Instance>) -> io::Result<()> {
let file = fs::File::open(find_workspace_path(
"tests-integration/resources/jsonbench-head-10.ndjson",
))?;
let reader = io::BufReader::new(file);
for line in reader.lines() {
for (i, line) in reader.lines().enumerate() {
let line = line?;
if line.is_empty() {
continue;
}
let json: serde_json::Value = serde_json::from_str(&line)?;
let time_us = json.pointer("/time_us").and_then(|x| x.as_u64()).unwrap();
let sql = format!(
"INSERT INTO bluesky (time_us, data) VALUES ({}, '{}')",
time_us,
"INSERT INTO bluesky (ts, data) VALUES ({}, '{}')",
i + 1,
line.replace("'", "''"), // standard method to escape the single quote
);
execute_sql_and_expect(frontend, &sql, "Affected Rows: 1").await;
@@ -300,12 +197,12 @@ async fn insert_data_by_sql(frontend: &Arc<Instance>) -> io::Result<()> {
async fn desc_table(frontend: &Arc<Instance>) {
let sql = "DESC TABLE bluesky";
let expected = r#"
+---------+------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| Column | Type | Key | Null | Default | Semantic Type |
+---------+------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| data | Json<{"_raw":"<String>","commit.collection":"<String>","commit.operation":"<String>","did":"<String>","kind":"<String>","time_us":"<Number>"}> | | YES | | FIELD |
| time_us | TimestampMicrosecond | PRI | NO | | TIMESTAMP |
+---------+------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+"#;
+--------+----------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| Column | Type | Key | Null | Default | Semantic Type |
+--------+----------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| data | Json<Object{"_raw": String, "commit.collection": String, "commit.operation": String, "did": String, "kind": String, "time_us": Number(I64)}> | | YES | | FIELD |
| ts | TimestampMillisecond | PRI | NO | | TIMESTAMP |
+--------+----------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+"#;
execute_sql_and_expect(frontend, sql, expected).await;
}
@@ -322,7 +219,7 @@ CREATE TABLE bluesky (
time_us Bigint
>,
),
time_us TimestampMicrosecond TIME INDEX,
ts Timestamp TIME INDEX,
)
"#;
execute_sql_and_expect(frontend, sql, "Affected Rows: 0").await;

View File

@@ -32,17 +32,6 @@ SHOW CREATE TABLE comment_table_test;
| | ) |
+--------------------+---------------------------------------------------+
SELECT table_comment
FROM information_schema.tables
WHERE table_schema = 'public'
AND table_name = 'comment_table_test';
+-------------------------+
| table_comment |
+-------------------------+
| table level description |
+-------------------------+
-- Remove table comment
COMMENT ON TABLE comment_table_test IS NULL;
@@ -65,17 +54,6 @@ SHOW CREATE TABLE comment_table_test;
| | |
+--------------------+---------------------------------------------------+
SELECT table_comment
FROM information_schema.tables
WHERE table_schema = 'public'
AND table_name = 'comment_table_test';
+---------------+
| table_comment |
+---------------+
| |
+---------------+
DROP TABLE comment_table_test;
Affected Rows: 0
@@ -112,18 +90,6 @@ SHOW CREATE TABLE comment_column_test;
| | |
+---------------------+---------------------------------------------------------+
SELECT column_comment
FROM information_schema.columns
WHERE table_schema = 'public'
AND table_name = 'comment_column_test'
AND column_name = 'val';
+--------------------------+
| column_comment |
+--------------------------+
| value column description |
+--------------------------+
-- Remove column comment
COMMENT ON COLUMN comment_column_test.val IS NULL;
@@ -146,18 +112,6 @@ SHOW CREATE TABLE comment_column_test;
| | |
+---------------------+----------------------------------------------------+
SELECT column_comment
FROM information_schema.columns
WHERE table_schema = 'public'
AND table_name = 'comment_column_test'
AND column_name = 'val';
+----------------+
| column_comment |
+----------------+
| |
+----------------+
DROP TABLE comment_column_test;
Affected Rows: 0
@@ -201,16 +155,6 @@ SHOW CREATE FLOW flow_comment_test;
| | AS SELECT desc_str, ts FROM flow_source_comment_test |
+-------------------+------------------------------------------------------+
SELECT comment
FROM information_schema.flows
WHERE flow_name = 'flow_comment_test';
+------------------------+
| comment |
+------------------------+
| flow level description |
+------------------------+
-- Remove flow comment
COMMENT ON FLOW flow_comment_test IS NULL;
@@ -226,16 +170,6 @@ SHOW CREATE FLOW flow_comment_test;
| | AS SELECT desc_str, ts FROM flow_source_comment_test |
+-------------------+------------------------------------------------------+
SELECT comment
FROM information_schema.flows
WHERE flow_name = 'flow_comment_test';
+---------+
| comment |
+---------+
| |
+---------+
DROP FLOW flow_comment_test;
Affected Rows: 0

View File

@@ -9,18 +9,10 @@ CREATE TABLE comment_table_test (
-- Add table comment
COMMENT ON TABLE comment_table_test IS 'table level description';
SHOW CREATE TABLE comment_table_test;
SELECT table_comment
FROM information_schema.tables
WHERE table_schema = 'public'
AND table_name = 'comment_table_test';
-- Remove table comment
COMMENT ON TABLE comment_table_test IS NULL;
SHOW CREATE TABLE comment_table_test;
SELECT table_comment
FROM information_schema.tables
WHERE table_schema = 'public'
AND table_name = 'comment_table_test';
DROP TABLE comment_table_test;
@@ -35,20 +27,10 @@ CREATE TABLE comment_column_test (
-- Add column comment
COMMENT ON COLUMN comment_column_test.val IS 'value column description';
SHOW CREATE TABLE comment_column_test;
SELECT column_comment
FROM information_schema.columns
WHERE table_schema = 'public'
AND table_name = 'comment_column_test'
AND column_name = 'val';
-- Remove column comment
COMMENT ON COLUMN comment_column_test.val IS NULL;
SHOW CREATE TABLE comment_column_test;
SELECT column_comment
FROM information_schema.columns
WHERE table_schema = 'public'
AND table_name = 'comment_column_test'
AND column_name = 'val';
DROP TABLE comment_column_test;
@@ -72,17 +54,12 @@ SELECT desc_str, ts FROM flow_source_comment_test;
-- Add flow comment
COMMENT ON FLOW flow_comment_test IS 'flow level description';
SHOW CREATE FLOW flow_comment_test;
SELECT comment
FROM information_schema.flows
WHERE flow_name = 'flow_comment_test';
-- Remove flow comment
COMMENT ON FLOW flow_comment_test IS NULL;
SHOW CREATE FLOW flow_comment_test;
SELECT comment
FROM information_schema.flows
WHERE flow_name = 'flow_comment_test';
DROP FLOW flow_comment_test;
DROP TABLE flow_source_comment_test;
DROP TABLE flow_sink_comment_test;

View File

@@ -12,7 +12,7 @@ DESC TABLE t;
| Column | Type | Key | Null | Default | Semantic Type |
+--------+----------------------+-----+------+---------+---------------+
| ts | TimestampMillisecond | PRI | NO | | TIMESTAMP |
| j | Json<"<Null>"> | | YES | | FIELD |
| j | Json<Null> | | YES | | FIELD |
+--------+----------------------+-----+------+---------+---------------+
INSERT INTO t VALUES
@@ -24,12 +24,12 @@ Affected Rows: 3
DESC TABLE t;
+--------+---------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| Column | Type | Key | Null | Default | Semantic Type |
+--------+---------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| ts | TimestampMillisecond | PRI | NO | | TIMESTAMP |
| j | Json<{"int":"<Number>","list":["<Number>"],"nested":{"a":{"x":"<String>"},"b":{"y":"<Number>"}}}> | | YES | | FIELD |
+--------+---------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
+--------+-----------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| Column | Type | Key | Null | Default | Semantic Type |
+--------+-----------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| ts | TimestampMillisecond | PRI | NO | | TIMESTAMP |
| j | Json<Object{"int": Number(I64), "list": Array[Number(F64)], "nested": Object{"a": Object{"x": String}, "b": Object{"y": Number(I64)}}}> | | YES | | FIELD |
+--------+-----------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
INSERT INTO t VALUES
(1762128004000, '{"int": 4, "bool": true, "nested": {"a": {"y": 1}}}'),
@@ -39,12 +39,12 @@ Affected Rows: 2
DESC TABLE t;
+--------+-------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| Column | Type | Key | Null | Default | Semantic Type |
+--------+-------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| ts | TimestampMillisecond | PRI | NO | | TIMESTAMP |
| j | Json<{"bool":"<Bool>","int":"<Number>","list":["<Number>"],"nested":{"a":{"x":"<String>","y":"<Number>"},"b":{"x":"<String>","y":"<Number>"}}}> | | YES | | FIELD |
+--------+-------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
+--------+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| Column | Type | Key | Null | Default | Semantic Type |
+--------+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| ts | TimestampMillisecond | PRI | NO | | TIMESTAMP |
| j | Json<Object{"bool": Bool, "int": Number(I64), "list": Array[Number(F64)], "nested": Object{"a": Object{"x": String, "y": Number(I64)}, "b": Object{"x": String, "y": Number(I64)}}}> | | YES | | FIELD |
+--------+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
INSERT INTO t VALUES (1762128006000, '{"int": 6, "list": [-6.0], "bool": true, "nested": {"a": {"x": "ax", "y": 66}, "b": {"y": -66, "x": "bx"}}}');
@@ -52,12 +52,12 @@ Affected Rows: 1
DESC TABLE t;
+--------+-------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| Column | Type | Key | Null | Default | Semantic Type |
+--------+-------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| ts | TimestampMillisecond | PRI | NO | | TIMESTAMP |
| j | Json<{"bool":"<Bool>","int":"<Number>","list":["<Number>"],"nested":{"a":{"x":"<String>","y":"<Number>"},"b":{"x":"<String>","y":"<Number>"}}}> | | YES | | FIELD |
+--------+-------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
+--------+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| Column | Type | Key | Null | Default | Semantic Type |
+--------+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
| ts | TimestampMillisecond | PRI | NO | | TIMESTAMP |
| j | Json<Object{"bool": Bool, "int": Number(I64), "list": Array[Number(F64)], "nested": Object{"a": Object{"x": String, "y": Number(I64)}, "b": Object{"x": String, "y": Number(I64)}}}> | | YES | | FIELD |
+--------+--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-----+------+---------+---------------+
INSERT INTO t VALUES (1762128011000, '{}');