Compare commits

..

2 Commits

Author SHA1 Message Date
John Spray
f5d83324bf update test 2025-04-08 17:56:30 +02:00
John Spray
7160787756 pageserver: add changed_bytes_from_parent consumption metric 2025-04-07 20:20:25 +02:00
18 changed files with 88 additions and 82 deletions

4
Cargo.lock generated
View File

@@ -7116,9 +7116,9 @@ dependencies = [
[[package]]
name = "tokio"
version = "1.43.1"
version = "1.43.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "492a604e2fd7f814268a378409e6c92b5525d747d10db9a229723f55a417958c"
checksum = "3d61fa4ffa3de412bfea335c6ecff681de2b609ba3c77ef3e00e521813a9ed9e"
dependencies = [
"backtrace",
"bytes",

View File

@@ -183,7 +183,7 @@ test-context = "0.3"
thiserror = "1.0"
tikv-jemallocator = { version = "0.6", features = ["profiling", "stats", "unprefixed_malloc_on_supported_platforms"] }
tikv-jemalloc-ctl = { version = "0.6", features = ["stats"] }
tokio = { version = "1.43.1", features = ["macros"] }
tokio = { version = "1.41", features = ["macros"] }
tokio-epoll-uring = { git = "https://github.com/neondatabase/tokio-epoll-uring.git" , branch = "main" }
tokio-io-timeout = "1.2.0"
tokio-postgres-rustls = "0.12.0"

View File

@@ -1527,51 +1527,6 @@ COPY --from=pgauditlogtofile-src /ext-src/ /ext-src/
WORKDIR /ext-src/pgauditlogtofile-src
RUN make install USE_PGXS=1 -j $(getconf _NPROCESSORS_ONLN)
#########################################################################################
#
# Layer "pg_rest-build"
# compile pg_rest extension
#
#########################################################################################
FROM build-deps AS pg_rest-src
ARG PG_VERSION
ARG PG_REST_VERSION=3.0.1
# Only supported for PostgreSQL v17
RUN if [ "${PG_VERSION:?}" != "v17" ]; then \
echo "pg_rest extension is only supported for PostgreSQL v17" && exit 0; \
fi
WORKDIR /ext-src
RUN mkdir -p pg_rest-src && cd pg_rest-src && \
wget https://github.com/ruslantalpa/foxfirebase/raw/main/pg_rest_pg17-${PG_REST_VERSION}_neon-debian-bookworm_aarch64.deb -O pg_rest_pg17-${PG_REST_VERSION}_aarch64.deb && \
wget https://github.com/ruslantalpa/foxfirebase/raw/main/pg_rest_pg17-${PG_REST_VERSION}_neon-debian-bookworm_amd64.deb -O pg_rest_pg17-${PG_REST_VERSION}_amd64.deb
FROM pg-build AS pg_rest-build
ARG PG_REST_VERSION=3.0.1
ARG PG_VERSION
COPY --from=pg_rest-src /ext-src/ /ext-src/
WORKDIR /ext-src/pg_rest-src
RUN if [ "${PG_VERSION:?}" = "v17" ]; then \
export ARCH=$(uname -m) && \
echo "ARCH: $ARCH" && \
# Map architecture names to package names
if [ "$ARCH" = "x86_64" ]; then \
PACKAGE_ARCH="amd64"; \
else \
PACKAGE_ARCH="$ARCH"; \
fi && \
echo "Using package architecture: $PACKAGE_ARCH" && \
apt-get update && \
apt-get install -y ./pg_rest_pg17-${PG_REST_VERSION}_${PACKAGE_ARCH}.deb && \
sed -i 's/superuser = false/superuser = true/g' /usr/local/pgsql/share/extension/pg_rest.control && \
echo 'trusted = true' >> /usr/local/pgsql/share/extension/pg_rest.control && \
# Clean up
apt-get clean && rm -rf /var/lib/apt/lists/*; \
else \
echo "pg_rest extension is only supported for PostgreSQL v17, skipping build"; \
fi
#########################################################################################
#
# Layer "neon-ext-build"
@@ -1667,7 +1622,6 @@ COPY --from=pg_duckdb-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_repack-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pgaudit-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pgauditlogtofile-build /usr/local/pgsql/ /usr/local/pgsql/
COPY --from=pg_rest-build /usr/local/pgsql/ /usr/local/pgsql/
#########################################################################################
#
@@ -1844,6 +1798,7 @@ COPY --from=pg_repack-src /ext-src/ /ext-src/
COPY --from=pg_repack-build /usr/local/pgsql/ /usr/local/pgsql/
COPY compute/patches/pg_repack.patch /ext-src
RUN cd /ext-src/pg_repack-src && patch -p1 </ext-src/pg_repack.patch && rm -f /ext-src/pg_repack.patch
COPY --chmod=755 docker-compose/run-tests.sh /run-tests.sh
RUN apt-get update && apt-get install -y libtap-parser-sourcehandler-pgtap-perl\
&& apt clean && rm -rf /ext-src/*.tar.gz /var/lib/apt/lists/*
@@ -1976,4 +1931,3 @@ RUN mkdir /var/run/rsyslogd && \
ENV LANG=en_US.utf8
USER postgres
ENTRYPOINT ["/usr/local/bin/compute_ctl"]

View File

@@ -27,6 +27,9 @@ pub(super) enum Name {
/// Timeline logical size
#[serde(rename = "timeline_logical_size")]
LogicalSize,
/// Timeline delta from parent
#[serde(rename = "timeline_changed_bytes_from_parent")]
ChangesFromParent,
/// Tenant remote size
#[serde(rename = "remote_storage_size")]
RemoteSize,
@@ -175,6 +178,18 @@ impl MetricsKey {
.absolute_values()
}
const fn timeline_changed_bytes_from_parent(
tenant_id: TenantId,
timeline_id: TimelineId,
) -> AbsoluteValueFactory {
MetricsKey {
tenant_id,
timeline_id: Some(timeline_id),
metric: Name::ChangesFromParent,
}
.absolute_values()
}
/// [`Tenant::remote_size`]
///
/// [`Tenant::remote_size`]: crate::tenant::Tenant::remote_size
@@ -364,6 +379,7 @@ struct TimelineSnapshot {
loaded_at: (Lsn, SystemTime),
last_record_lsn: Lsn,
current_exact_logical_size: Option<u64>,
changed_bytes_from_parent: Option<u64>,
}
impl TimelineSnapshot {
@@ -399,10 +415,23 @@ impl TimelineSnapshot {
}
};
// This is an approximation of how much data has changed on this branch vs. its ancestor: the
// number of bytes written to the WAL, clamped to the size of the branch.
let changed_bytes_from_parent = current_exact_logical_size.and_then(|size| {
if t.get_ancestor_lsn() == Lsn::MAX {
None
} else {
t.get_last_record_lsn()
.checked_sub(t.get_ancestor_lsn())
.map(|wal_bytes| std::cmp::min(wal_bytes.0, size))
}
});
Ok(Some(TimelineSnapshot {
loaded_at,
last_record_lsn,
current_exact_logical_size,
changed_bytes_from_parent,
}))
}
}
@@ -480,6 +509,17 @@ impl TimelineSnapshot {
metrics.push(factory.at(now, size));
}
}
{
let factory = MetricsKey::timeline_changed_bytes_from_parent(tenant_id, timeline_id);
let current_or_previous = self
.changed_bytes_from_parent
.or_else(|| cache.get(factory.key()).map(|item| item.value));
if let Some(size) = current_or_previous {
metrics.push(factory.at(now, size));
}
}
}
}

View File

@@ -18,6 +18,7 @@ fn startup_collected_timeline_metrics_before_advancing() {
loaded_at: (disk_consistent_lsn, SystemTime::now()),
last_record_lsn: disk_consistent_lsn,
current_exact_logical_size: Some(0x42000),
changed_bytes_from_parent: Some(0x1000),
};
let now = DateTime::<Utc>::from(SystemTime::now());
@@ -33,7 +34,8 @@ fn startup_collected_timeline_metrics_before_advancing() {
0
),
MetricsKey::written_size(tenant_id, timeline_id).at(now, disk_consistent_lsn.0),
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0x42000)
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0x42000),
MetricsKey::timeline_changed_bytes_from_parent(tenant_id, timeline_id).at(now, 0x1000)
]
);
}
@@ -60,6 +62,7 @@ fn startup_collected_timeline_metrics_second_round() {
loaded_at: (disk_consistent_lsn, init),
last_record_lsn: disk_consistent_lsn,
current_exact_logical_size: Some(0x42000),
changed_bytes_from_parent: Some(0x1000),
};
snap.to_metrics(tenant_id, timeline_id, now, &mut metrics, &cache);
@@ -69,7 +72,8 @@ fn startup_collected_timeline_metrics_second_round() {
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(before, now, 0),
MetricsKey::written_size(tenant_id, timeline_id).at(now, disk_consistent_lsn.0),
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0x42000)
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0x42000),
MetricsKey::timeline_changed_bytes_from_parent(tenant_id, timeline_id).at(now, 0x1000)
]
);
}
@@ -104,6 +108,7 @@ fn startup_collected_timeline_metrics_nth_round_at_same_lsn() {
loaded_at: (disk_consistent_lsn, init),
last_record_lsn: disk_consistent_lsn,
current_exact_logical_size: Some(0x42000),
changed_bytes_from_parent: Some(0x1000),
};
snap.to_metrics(tenant_id, timeline_id, now, &mut metrics, &cache);
@@ -113,7 +118,8 @@ fn startup_collected_timeline_metrics_nth_round_at_same_lsn() {
&[
MetricsKey::written_size_delta(tenant_id, timeline_id).from_until(just_before, now, 0),
MetricsKey::written_size(tenant_id, timeline_id).at(now, disk_consistent_lsn.0),
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0x42000)
MetricsKey::timeline_logical_size(tenant_id, timeline_id).at(now, 0x42000),
MetricsKey::timeline_changed_bytes_from_parent(tenant_id, timeline_id).at(now, 0x1000)
]
);
}
@@ -141,6 +147,7 @@ fn post_restart_written_sizes_with_rolled_back_last_record_lsn() {
loaded_at: (Lsn(50), at_restart),
last_record_lsn: Lsn(50),
current_exact_logical_size: None,
changed_bytes_from_parent: None,
};
let mut cache = HashMap::from([
@@ -202,6 +209,7 @@ fn post_restart_current_exact_logical_size_uses_cached() {
loaded_at: (Lsn(50), at_restart),
last_record_lsn: Lsn(50),
current_exact_logical_size: None,
changed_bytes_from_parent: Some(0x1000),
};
let cache = HashMap::from([MetricsKey::timeline_logical_size(tenant_id, timeline_id)

View File

@@ -1244,10 +1244,6 @@ impl Timeline {
let mut replace_image_layers = Vec::new();
for layer in layers_to_rewrite {
if self.cancel.is_cancelled() {
return Err(CompactionError::ShuttingDown);
}
tracing::info!(layer=%layer, "Rewriting layer after shard split...");
let mut image_layer_writer = ImageLayerWriter::new(
self.conf,

View File

@@ -274,8 +274,13 @@ typedef struct
XLogRecPtr effective_request_lsn;
} neon_request_lsns;
#if PG_MAJORVERSION_NUM < 16
extern PGDLLEXPORT void neon_read_at_lsn(NRelFileInfo rnode, ForkNumber forkNum, BlockNumber blkno,
neon_request_lsns request_lsns, char *buffer);
#else
extern PGDLLEXPORT void neon_read_at_lsn(NRelFileInfo rnode, ForkNumber forkNum, BlockNumber blkno,
neon_request_lsns request_lsns, void *buffer);
#endif
extern int64 neon_dbsize(Oid dbNode);
/* utils for neon relsize cache */

View File

@@ -1900,6 +1900,7 @@ neon_wallog_pagev(SMgrRelation reln, ForkNumber forknum, BlockNumber blocknum,
log_pages = true;
}
else if (XLogInsertAllowed() &&
!ShutdownRequestPending &&
(forknum == FSM_FORKNUM || forknum == VISIBILITYMAP_FORKNUM))
{
log_pages = true;
@@ -3153,8 +3154,13 @@ neon_writeback(SMgrRelation reln, ForkNumber forknum,
* The offsets in request_lsns, buffers, and mask are linked.
*/
static void
#if PG_MAJORVERSION_NUM < 16
neon_read_at_lsnv(NRelFileInfo rinfo, ForkNumber forkNum, BlockNumber base_blockno, neon_request_lsns *request_lsns,
char **buffers, BlockNumber nblocks, const bits8 *mask)
#else
neon_read_at_lsnv(NRelFileInfo rinfo, ForkNumber forkNum, BlockNumber base_blockno, neon_request_lsns *request_lsns,
void **buffers, BlockNumber nblocks, const bits8 *mask)
#endif
{
NeonResponse *resp;
uint64 ring_index;
@@ -3350,8 +3356,13 @@ Retry:
* To avoid breaking tests in the runtime please keep function signature in sync.
*/
void
#if PG_MAJORVERSION_NUM < 16
neon_read_at_lsn(NRelFileInfo rinfo, ForkNumber forkNum, BlockNumber blkno,
neon_request_lsns request_lsns, char *buffer)
#else
neon_read_at_lsn(NRelFileInfo rinfo, ForkNumber forkNum, BlockNumber blkno,
neon_request_lsns request_lsns, void *buffer)
#endif
{
neon_read_at_lsnv(rinfo, forkNum, blkno, &request_lsns, &buffer, 1, NULL);
}

View File

@@ -50,8 +50,13 @@ PG_FUNCTION_INFO_V1(trigger_segfault);
* Linkage to functions in neon module.
* The signature here would need to be updated whenever function parameters change in pagestore_smgr.c
*/
#if PG_MAJORVERSION_NUM < 16
typedef void (*neon_read_at_lsn_type) (NRelFileInfo rinfo, ForkNumber forkNum, BlockNumber blkno,
neon_request_lsns request_lsns, char *buffer);
#else
typedef void (*neon_read_at_lsn_type) (NRelFileInfo rinfo, ForkNumber forkNum, BlockNumber blkno,
neon_request_lsns request_lsns, void *buffer);
#endif
static neon_read_at_lsn_type neon_read_at_lsn_ptr;

View File

@@ -223,9 +223,6 @@ struct Args {
/// Flag to use https for requests to peer's safekeeper API.
#[arg(long)]
pub use_https_safekeeper_api: bool,
/// Path to the JWT auth token used to authenticate with other safekeepers.
#[arg(long)]
auth_token_path: Option<Utf8PathBuf>,
}
// Like PathBufValueParser, but allows empty string.
@@ -344,24 +341,14 @@ async fn main() -> anyhow::Result<()> {
};
// Load JWT auth token to connect to other safekeepers for pull_timeline.
// First check if the env var is present, then check the arg with the path.
// We want to deprecate and remove the env var method in the future.
let sk_auth_token = match var("SAFEKEEPER_AUTH_TOKEN") {
Ok(v) => {
info!("loaded JWT token for authentication with safekeepers");
Some(SecretString::from(v))
}
Err(VarError::NotPresent) => {
if let Some(auth_token_path) = args.auth_token_path.as_ref() {
info!(
"loading JWT token for authentication with safekeepers from {auth_token_path}"
);
let auth_token = tokio::fs::read_to_string(auth_token_path).await?;
Some(SecretString::from(auth_token.trim().to_owned()))
} else {
info!("no JWT token for authentication with safekeepers detected");
None
}
info!("no JWT token for authentication with safekeepers detected");
None
}
Err(_) => {
warn!("JWT token for authentication with safekeepers is not unicode");

View File

@@ -899,7 +899,7 @@ async fn handle_node_status(req: Request<Body>) -> Result<Response<Body>, ApiErr
let state = get_state(&req);
let node_id: NodeId = parse_request_param(&req, "node_id")?;
let node_status = state.service.get_node(node_id).await?.describe();
let node_status = state.service.get_node(node_id).await?;
json_response(StatusCode::OK, node_status)
}

View File

@@ -511,4 +511,5 @@ PER_METRIC_VERIFIERS = {
"written_data_bytes_delta": WrittenDataDeltaVerifier,
"timeline_logical_size": CannotVerifyAnything,
"synthetic_storage_size": SyntheticSizeVerifier,
"timeline_changed_bytes_from_parent": CannotVerifyAnything,
}

View File

@@ -4109,7 +4109,6 @@ def test_storcon_create_delete_sk_down(neon_env_builder: NeonEnvBuilder, restart
env.storage_controller.allowed_errors.extend(
[
".*Call to safekeeper.* management API still failed after.*",
".*Call to safekeeper.* management API failed, will retry.*",
".*reconcile_one.*tenant_id={tenant_id}.*Call to safekeeper.* management API still failed after.*",
]
)

View File

@@ -1,18 +1,18 @@
{
"v17": [
"17.4",
"66114c23bc61205b0e3fb1e77ee76a4abc1eb4b8"
"c9e4ff5a38907acd71107634055bf2609aba43a5"
],
"v16": [
"16.8",
"d56e79cd5d6136c159b1d8d98acb7981d4b69364"
"746bd9ffe5c29bce030eaea1031054057f3c5d45"
],
"v15": [
"15.12",
"aeb292eeace9072e07071254b6ffc7a74007d4d2"
"23708b3aca9adf163aa0973eb63d9afc0e4a04c3"
],
"v14": [
"14.17",
"a0391901a2af13aa029b905272a5b2024133c926"
"8cca70c22e2894dd4645f9a940086ac437b0a11b"
]
}