safekeeper: fan out from single wal reader to multiple shards (#10190)

## Problem

Safekeepers currently decode and interpret WAL for each shard
separately.
This is wasteful in terms of CPU memory usage - we've seen this in
profiles.

## Summary of changes

Fan-out interpreted WAL to multiple shards. 
The basic is that wal decoding and interpretation happens in a separate
tokio task and senders
attach to it. Senders only receive batches concerning their shard and
only past the Lsn they've last seen.

Fan-out is gated behind the `wal_reader_fanout` safekeeper flag
(disabled by default for now).

When fan-out is enabled, it might be desirable to control the absolute
delta between the
current position and a new shard's desired position (i.e. how far behind
or ahead a shard may be).
`max_delta_for_fanout` is a new optional safekeeper flag which dictates
whether to create a new
WAL reader or attach to the existing one. By default, this behaviour is
disabled. Let's consider enabling
it if we spot the need for it in the field.

## Testing

Tests passed [here](https://github.com/neondatabase/neon/pull/10301)
with wal reader fanout enabled
as of
34f6a71718.

Related: https://github.com/neondatabase/neon/issues/9337
Epic: https://github.com/neondatabase/neon/issues/9329
This commit is contained in:
Vlad Lazar
2025-01-15 15:33:54 +00:00
committed by GitHub
parent 3e529f124f
commit dbebede7bf
16 changed files with 1410 additions and 270 deletions

View File

@@ -64,7 +64,7 @@ pub struct InterpretedWalRecords {
}
/// An interpreted Postgres WAL record, ready to be handled by the pageserver
#[derive(Serialize, Deserialize)]
#[derive(Serialize, Deserialize, Clone)]
pub struct InterpretedWalRecord {
/// Optional metadata record - may cause writes to metadata keys
/// in the storage engine

View File

@@ -32,7 +32,7 @@ static ZERO_PAGE: Bytes = Bytes::from_static(&[0u8; BLCKSZ as usize]);
/// relation sizes. In the case of "observed" values, we only need to know
/// the key and LSN, so two types of metadata are supported to save on network
/// bandwidth.
#[derive(Serialize, Deserialize)]
#[derive(Serialize, Deserialize, Clone)]
pub enum ValueMeta {
Serialized(SerializedValueMeta),
Observed(ObservedValueMeta),
@@ -79,7 +79,7 @@ impl PartialEq for OrderedValueMeta {
impl Eq for OrderedValueMeta {}
/// Metadata for a [`Value`] serialized into the batch.
#[derive(Serialize, Deserialize)]
#[derive(Serialize, Deserialize, Clone)]
pub struct SerializedValueMeta {
pub key: CompactKey,
pub lsn: Lsn,
@@ -91,14 +91,14 @@ pub struct SerializedValueMeta {
}
/// Metadata for a [`Value`] observed by the batch
#[derive(Serialize, Deserialize)]
#[derive(Serialize, Deserialize, Clone)]
pub struct ObservedValueMeta {
pub key: CompactKey,
pub lsn: Lsn,
}
/// Batch of serialized [`Value`]s.
#[derive(Serialize, Deserialize)]
#[derive(Serialize, Deserialize, Clone)]
pub struct SerializedValueBatch {
/// [`Value`]s serialized in EphemeralFile's native format,
/// ready for disk write by the pageserver