Files
neon/pageserver/src/repository.rs
Christian Schwarz 4825b0fec3 compaction_level0_phase1: bypass PS PageCache for data blocks (#8543)
part of https://github.com/neondatabase/neon/issues/8184

# Problem

We want to bypass PS PageCache for all data block reads, but
`compact_level0_phase1` currently uses `ValueRef::load` to load the WAL
records from delta layers.
Internally, that maps to `FileBlockReader:read_blk` which hits the
PageCache
[here](e78341e1c2/pageserver/src/tenant/block_io.rs (L229-L236)).

# Solution

This PR adds a mode for `compact_level0_phase1` that uses the
`MergeIterator` for reading the `Value`s from the delta layer files.

`MergeIterator` is a streaming k-merge that uses vectored blob_io under
the hood, which bypasses the PS PageCache for data blocks.

Other notable changes:
* change the `DiskBtreeReader::into_stream` to buffer the node, instead
of holding a `PageCache` `PageReadGuard`.
* Without this, we run out of page cache slots in
`test_pageserver_compaction_smoke`.
* Generally, `PageReadGuard`s aren't supposed to be held across await
points, so, this is a general bugfix.

# Testing / Validation / Performance

`MergeIterator` has not yet been used in production; it's being
developed as part of
* https://github.com/neondatabase/neon/issues/8002

Therefore, this PR adds a validation mode that compares the existing
approach's value iterator with the new approach's stream output, item by
item.
If they're not identical, we log a warning / fail the unit/regression
test.
To avoid flooding the logs, we apply a global rate limit of once per 10
seconds.
In any case, we use the existing approach's value.

Expected performance impact that will be monitored in staging / nightly
benchmarks / eventually pre-prod:
* with validation:
  * increased CPU usage
  * ~doubled VirtualFile read bytes/second metric
* no change in disk IO usage because the kernel page cache will likely
have the pages buffered on the second read
* without validation:
* slightly higher DRAM usage because each iterator participating in the
k-merge has a dedicated buffer (as opposed to before, where compactions
would rely on the PS PageCaceh as a shared evicting buffer)
* less disk IO if previously there were repeat PageCache misses (likely
case on a busy production Pageserver)
* lower CPU usage: PageCache out of the picture, fewer syscalls are made
(vectored blob io batches reads)

# Rollout

The new code is used with validation mode enabled-by-default.
This gets us validation everywhere by default, specifically in
- Rust unit tests
- Python tests
- Nightly pagebench (shouldn't really matter)
- Staging

Before the next release, I'll merge the following aws.git PR that
configures prod to continue using the existing behavior:

* https://github.com/neondatabase/aws/pull/1663

# Interactions With Other Features

This work & rollout should complete before Direct IO is enabled because
Direct IO would double the IOPS & latency for each compaction read
(#8240).

# Future Work

The streaming k-merge's memory usage is proportional to the amount of
memory per participating layer.

But `compact_level0_phase1` still loads all keys into memory for
`all_keys_iter`.
Thus, it continues to have active memory usage proportional to the
number of keys involved in the compaction.

Future work should replace `all_keys_iter` with a streaming keys
iterator.
This PR has a draft in its first commit, which I later reverted because
it's not necessary to achieve the goal of this PR / issue #8184.
2024-07-31 14:17:59 +02:00

285 lines
7.9 KiB
Rust

use crate::walrecord::NeonWalRecord;
use anyhow::Result;
use bytes::Bytes;
use serde::{Deserialize, Serialize};
use std::ops::AddAssign;
use std::time::Duration;
pub use pageserver_api::key::{Key, KEY_SIZE};
/// A 'value' stored for a one Key.
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
pub enum Value {
/// An Image value contains a full copy of the value
Image(Bytes),
/// A WalRecord value contains a WAL record that needs to be
/// replayed get the full value. Replaying the WAL record
/// might need a previous version of the value (if will_init()
/// returns false), or it may be replayed stand-alone (true).
WalRecord(NeonWalRecord),
}
impl Value {
pub fn is_image(&self) -> bool {
matches!(self, Value::Image(_))
}
pub fn will_init(&self) -> bool {
match self {
Value::Image(_) => true,
Value::WalRecord(rec) => rec.will_init(),
}
}
}
#[derive(Debug, PartialEq)]
pub(crate) enum InvalidInput {
TooShortValue,
TooShortPostgresRecord,
}
/// We could have a ValueRef where everything is `serde(borrow)`. Before implementing that, lets
/// use this type for querying if a slice looks some particular way.
pub(crate) struct ValueBytes;
impl ValueBytes {
pub(crate) fn will_init(raw: &[u8]) -> Result<bool, InvalidInput> {
if raw.len() < 12 {
return Err(InvalidInput::TooShortValue);
}
let value_discriminator = &raw[0..4];
if value_discriminator == [0, 0, 0, 0] {
// Value::Image always initializes
return Ok(true);
}
if value_discriminator != [0, 0, 0, 1] {
// not a Value::WalRecord(..)
return Ok(false);
}
let walrecord_discriminator = &raw[4..8];
if walrecord_discriminator != [0, 0, 0, 0] {
// only NeonWalRecord::Postgres can have will_init
return Ok(false);
}
if raw.len() < 17 {
return Err(InvalidInput::TooShortPostgresRecord);
}
Ok(raw[8] == 1)
}
}
#[cfg(test)]
mod test {
use super::*;
use utils::bin_ser::BeSer;
macro_rules! roundtrip {
($orig:expr, $expected:expr) => {{
let orig: Value = $orig;
let actual = Value::ser(&orig).unwrap();
let expected: &[u8] = &$expected;
assert_eq!(utils::Hex(&actual), utils::Hex(expected));
let deser = Value::des(&actual).unwrap();
assert_eq!(orig, deser);
}};
}
#[test]
fn image_roundtrip() {
let image = Bytes::from_static(b"foobar");
let image = Value::Image(image);
#[rustfmt::skip]
let expected = [
// top level discriminator of 4 bytes
0x00, 0x00, 0x00, 0x00,
// 8 byte length
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06,
// foobar
0x66, 0x6f, 0x6f, 0x62, 0x61, 0x72
];
roundtrip!(image, expected);
assert!(ValueBytes::will_init(&expected).unwrap());
}
#[test]
fn walrecord_postgres_roundtrip() {
let rec = NeonWalRecord::Postgres {
will_init: true,
rec: Bytes::from_static(b"foobar"),
};
let rec = Value::WalRecord(rec);
#[rustfmt::skip]
let expected = [
// flattened discriminator of total 8 bytes
0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x00,
// will_init
0x01,
// 8 byte length
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x06,
// foobar
0x66, 0x6f, 0x6f, 0x62, 0x61, 0x72
];
roundtrip!(rec, expected);
assert!(ValueBytes::will_init(&expected).unwrap());
}
#[test]
fn bytes_inspection_too_short_image() {
let rec = Value::Image(Bytes::from_static(b""));
#[rustfmt::skip]
let expected = [
// top level discriminator of 4 bytes
0x00, 0x00, 0x00, 0x00,
// 8 byte length
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
];
roundtrip!(rec, expected);
assert!(ValueBytes::will_init(&expected).unwrap());
assert_eq!(expected.len(), 12);
for len in 0..12 {
assert_eq!(
ValueBytes::will_init(&expected[..len]).unwrap_err(),
InvalidInput::TooShortValue
);
}
}
#[test]
fn bytes_inspection_too_short_postgres_record() {
let rec = NeonWalRecord::Postgres {
will_init: false,
rec: Bytes::from_static(b""),
};
let rec = Value::WalRecord(rec);
#[rustfmt::skip]
let expected = [
// flattened discriminator of total 8 bytes
0x00, 0x00, 0x00, 0x01,
0x00, 0x00, 0x00, 0x00,
// will_init
0x00,
// 8 byte length
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
];
roundtrip!(rec, expected);
assert!(!ValueBytes::will_init(&expected).unwrap());
assert_eq!(expected.len(), 17);
for len in 12..17 {
assert_eq!(
ValueBytes::will_init(&expected[..len]).unwrap_err(),
InvalidInput::TooShortPostgresRecord
)
}
for len in 0..12 {
assert_eq!(
ValueBytes::will_init(&expected[..len]).unwrap_err(),
InvalidInput::TooShortValue
)
}
}
#[test]
fn clear_visibility_map_flags_example() {
let rec = NeonWalRecord::ClearVisibilityMapFlags {
new_heap_blkno: Some(0x11),
old_heap_blkno: None,
flags: 0x03,
};
let rec = Value::WalRecord(rec);
#[rustfmt::skip]
let expected = [
// discriminators
0x00, 0x00, 0x00, 0x01,
0x00, 0x00, 0x00, 0x01,
// Some == 1 followed by 4 bytes
0x01, 0x00, 0x00, 0x00, 0x11,
// None == 0
0x00,
// flags
0x03
];
roundtrip!(rec, expected);
assert!(!ValueBytes::will_init(&expected).unwrap());
}
}
///
/// Result of performing GC
///
#[derive(Default, Serialize, Debug)]
pub struct GcResult {
pub layers_total: u64,
pub layers_needed_by_cutoff: u64,
pub layers_needed_by_pitr: u64,
pub layers_needed_by_branches: u64,
pub layers_needed_by_leases: u64,
pub layers_not_updated: u64,
pub layers_removed: u64, // # of layer files removed because they have been made obsolete by newer ondisk files.
#[serde(serialize_with = "serialize_duration_as_millis")]
pub elapsed: Duration,
/// The layers which were garbage collected.
///
/// Used in `/v1/tenant/:tenant_id/timeline/:timeline_id/do_gc` to wait for the layers to be
/// dropped in tests.
#[cfg(feature = "testing")]
#[serde(skip)]
pub(crate) doomed_layers: Vec<crate::tenant::storage_layer::Layer>,
}
// helper function for `GcResult`, serializing a `Duration` as an integer number of milliseconds
fn serialize_duration_as_millis<S>(d: &Duration, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::Serializer,
{
d.as_millis().serialize(serializer)
}
impl AddAssign for GcResult {
fn add_assign(&mut self, other: Self) {
self.layers_total += other.layers_total;
self.layers_needed_by_pitr += other.layers_needed_by_pitr;
self.layers_needed_by_cutoff += other.layers_needed_by_cutoff;
self.layers_needed_by_branches += other.layers_needed_by_branches;
self.layers_needed_by_leases += other.layers_needed_by_leases;
self.layers_not_updated += other.layers_not_updated;
self.layers_removed += other.layers_removed;
self.elapsed += other.elapsed;
#[cfg(feature = "testing")]
{
let mut other = other;
self.doomed_layers.append(&mut other.doomed_layers);
}
}
}