pageserver: simplify LayerAccessStats (#8431)

## Problem

LayerAccessStats contains a lot of detail that we don't use: short
histories of most recent accesses, specifics on what kind of task
accessed a layer, etc. This is all stored inside a Mutex, which is
locked every time something accesses a layer.

## Summary of changes

- Store timestamps at a very low resolution (to the nearest second),
sufficient for use on the timescales of eviction.
- Pack access time and last residence change time into a single u64
- Use the high bits of the u64 for other flags, including the new layer
visibility concept.
- Simplify the external-facing model for access stats to just include
what we now track.

Note that the `HistoryBufferWithDropCounter` is removed here because it
is no longer used. I do not dislike this type, we just happen not to use
it for anything else at present.


Co-authored-by: Christian Schwarz <christian@neon.tech>
This commit is contained in:
John Spray
2024-07-24 08:17:28 +01:00
committed by GitHub
parent 925c5ad1e8
commit f5db655447
15 changed files with 209 additions and 579 deletions

View File

@@ -5,7 +5,6 @@ pub mod utilization;
pub use utilization::PageserverUtilization;
use std::{
borrow::Cow,
collections::HashMap,
io::{BufRead, Read},
num::{NonZeroU64, NonZeroUsize},
@@ -20,7 +19,6 @@ use serde::{Deserialize, Serialize};
use serde_with::serde_as;
use utils::{
completion,
history_buffer::HistoryBufferWithDropCounter,
id::{NodeId, TenantId, TimelineId},
lsn::Lsn,
serde_system_time,
@@ -726,58 +724,7 @@ pub struct LayerMapInfo {
pub historic_layers: Vec<HistoricLayerInfo>,
}
#[derive(Debug, Hash, PartialEq, Eq, Clone, Copy, Serialize, Deserialize, enum_map::Enum)]
#[repr(usize)]
pub enum LayerAccessKind {
GetValueReconstructData,
Iter,
KeyIter,
Dump,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct LayerAccessStatFullDetails {
pub when_millis_since_epoch: u64,
pub task_kind: Cow<'static, str>,
pub access_kind: LayerAccessKind,
}
/// An event that impacts the layer's residence status.
#[serde_as]
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct LayerResidenceEvent {
/// The time when the event occurred.
/// NB: this timestamp is captured while the residence status changes.
/// So, it might be behind/ahead of the actual residence change by a short amount of time.
///
#[serde(rename = "timestamp_millis_since_epoch")]
#[serde_as(as = "serde_with::TimestampMilliSeconds")]
pub timestamp: SystemTime,
/// The new residence status of the layer.
pub status: LayerResidenceStatus,
/// The reason why we had to record this event.
pub reason: LayerResidenceEventReason,
}
/// The reason for recording a given [`LayerResidenceEvent`].
#[derive(Debug, Clone, Copy, Serialize, Deserialize)]
pub enum LayerResidenceEventReason {
/// The layer map is being populated, e.g. during timeline load or attach.
/// This includes [`RemoteLayer`] objects created in [`reconcile_with_remote`].
/// We need to record such events because there is no persistent storage for the events.
///
// https://github.com/rust-lang/rust/issues/74481
/// [`RemoteLayer`]: ../../tenant/storage_layer/struct.RemoteLayer.html
/// [`reconcile_with_remote`]: ../../tenant/struct.Timeline.html#method.reconcile_with_remote
LayerLoad,
/// We just created the layer (e.g., freeze_and_flush or compaction).
/// Such layers are always [`LayerResidenceStatus::Resident`].
LayerCreate,
/// We on-demand downloaded or evicted the given layer.
ResidenceChange,
}
/// The residence status of the layer, after the given [`LayerResidenceEvent`].
/// The residence status of a layer
#[derive(Debug, Clone, Copy, Serialize, Deserialize)]
pub enum LayerResidenceStatus {
/// Residence status for a layer file that exists locally.
@@ -787,23 +734,16 @@ pub enum LayerResidenceStatus {
Evicted,
}
impl LayerResidenceEvent {
pub fn new(status: LayerResidenceStatus, reason: LayerResidenceEventReason) -> Self {
Self {
status,
reason,
timestamp: SystemTime::now(),
}
}
}
#[serde_as]
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct LayerAccessStats {
pub access_count_by_access_kind: HashMap<LayerAccessKind, u64>,
pub task_kind_access_flag: Vec<Cow<'static, str>>,
pub first: Option<LayerAccessStatFullDetails>,
pub accesses_history: HistoryBufferWithDropCounter<LayerAccessStatFullDetails, 16>,
pub residence_events_history: HistoryBufferWithDropCounter<LayerResidenceEvent, 16>,
#[serde_as(as = "serde_with::TimestampMilliSeconds")]
pub access_time: SystemTime,
#[serde_as(as = "serde_with::TimestampMilliSeconds")]
pub residence_time: SystemTime,
pub visible: bool,
}
#[derive(Debug, Clone, Serialize, Deserialize)]

View File

@@ -20,7 +20,6 @@ bincode.workspace = true
bytes.workspace = true
camino.workspace = true
chrono.workspace = true
heapless.workspace = true
hex = { workspace = true, features = ["serde"] }
humantime.workspace = true
hyper = { workspace = true, features = ["full"] }

View File

@@ -1,196 +0,0 @@
//! A heapless buffer for events of sorts.
use std::ops;
use heapless::HistoryBuffer;
#[derive(Debug, Clone)]
pub struct HistoryBufferWithDropCounter<T, const L: usize> {
buffer: HistoryBuffer<T, L>,
drop_count: u64,
}
impl<T, const L: usize> HistoryBufferWithDropCounter<T, L> {
pub fn write(&mut self, data: T) {
let len_before = self.buffer.len();
self.buffer.write(data);
let len_after = self.buffer.len();
self.drop_count += u64::from(len_before == len_after);
}
pub fn drop_count(&self) -> u64 {
self.drop_count
}
pub fn map<U, F: Fn(&T) -> U>(&self, f: F) -> HistoryBufferWithDropCounter<U, L> {
let mut buffer = HistoryBuffer::new();
buffer.extend(self.buffer.oldest_ordered().map(f));
HistoryBufferWithDropCounter::<U, L> {
buffer,
drop_count: self.drop_count,
}
}
}
impl<T, const L: usize> Default for HistoryBufferWithDropCounter<T, L> {
fn default() -> Self {
Self {
buffer: HistoryBuffer::default(),
drop_count: 0,
}
}
}
impl<T, const L: usize> ops::Deref for HistoryBufferWithDropCounter<T, L> {
type Target = HistoryBuffer<T, L>;
fn deref(&self) -> &Self::Target {
&self.buffer
}
}
#[derive(serde::Serialize, serde::Deserialize)]
struct SerdeRepr<T> {
buffer: Vec<T>,
buffer_size: usize,
drop_count: u64,
}
impl<'a, T, const L: usize> From<&'a HistoryBufferWithDropCounter<T, L>> for SerdeRepr<T>
where
T: Clone + serde::Serialize,
{
fn from(value: &'a HistoryBufferWithDropCounter<T, L>) -> Self {
let HistoryBufferWithDropCounter { buffer, drop_count } = value;
SerdeRepr {
buffer: buffer.iter().cloned().collect(),
buffer_size: L,
drop_count: *drop_count,
}
}
}
impl<T, const L: usize> serde::Serialize for HistoryBufferWithDropCounter<T, L>
where
T: Clone + serde::Serialize,
{
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::Serializer,
{
SerdeRepr::from(self).serialize(serializer)
}
}
impl<'de, T, const L: usize> serde::de::Deserialize<'de> for HistoryBufferWithDropCounter<T, L>
where
T: Clone + serde::Deserialize<'de>,
{
fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
where
D: serde::Deserializer<'de>,
{
let SerdeRepr {
buffer: des_buffer,
drop_count,
buffer_size,
} = SerdeRepr::<T>::deserialize(deserializer)?;
if buffer_size != L {
use serde::de::Error;
return Err(D::Error::custom(format!(
"invalid buffer_size, expecting {L} got {buffer_size}"
)));
}
let mut buffer = HistoryBuffer::new();
buffer.extend(des_buffer);
Ok(HistoryBufferWithDropCounter { buffer, drop_count })
}
}
#[cfg(test)]
mod test {
use super::HistoryBufferWithDropCounter;
#[test]
fn test_basics() {
let mut b = HistoryBufferWithDropCounter::<usize, 2>::default();
b.write(1);
b.write(2);
b.write(3);
assert!(b.iter().any(|e| *e == 2));
assert!(b.iter().any(|e| *e == 3));
assert!(!b.iter().any(|e| *e == 1));
// round-trip serde
let round_tripped: HistoryBufferWithDropCounter<usize, 2> =
serde_json::from_str(&serde_json::to_string(&b).unwrap()).unwrap();
assert_eq!(
round_tripped.iter().cloned().collect::<Vec<_>>(),
b.iter().cloned().collect::<Vec<_>>()
);
}
#[test]
fn test_drop_count_works() {
let mut b = HistoryBufferWithDropCounter::<_, 2>::default();
b.write(1);
assert_eq!(b.drop_count(), 0);
b.write(2);
assert_eq!(b.drop_count(), 0);
b.write(3);
assert_eq!(b.drop_count(), 1);
b.write(4);
assert_eq!(b.drop_count(), 2);
}
#[test]
fn test_clone_works() {
let mut b = HistoryBufferWithDropCounter::<_, 2>::default();
b.write(1);
b.write(2);
b.write(3);
assert_eq!(b.drop_count(), 1);
let mut c = b.clone();
assert_eq!(c.drop_count(), 1);
assert!(c.iter().any(|e| *e == 2));
assert!(c.iter().any(|e| *e == 3));
assert!(!c.iter().any(|e| *e == 1));
c.write(4);
assert!(c.iter().any(|e| *e == 4));
assert!(!b.iter().any(|e| *e == 4));
}
#[test]
fn test_map() {
let mut b = HistoryBufferWithDropCounter::<_, 2>::default();
b.write(1);
assert_eq!(b.drop_count(), 0);
{
let c = b.map(|i| i + 10);
assert_eq!(c.oldest_ordered().cloned().collect::<Vec<_>>(), vec![11]);
assert_eq!(c.drop_count(), 0);
}
b.write(2);
assert_eq!(b.drop_count(), 0);
{
let c = b.map(|i| i + 10);
assert_eq!(
c.oldest_ordered().cloned().collect::<Vec<_>>(),
vec![11, 12]
);
assert_eq!(c.drop_count(), 0);
}
b.write(3);
assert_eq!(b.drop_count(), 1);
{
let c = b.map(|i| i + 10);
assert_eq!(
c.oldest_ordered().cloned().collect::<Vec<_>>(),
vec![12, 13]
);
assert_eq!(c.drop_count(), 1);
}
}
}

View File

@@ -59,8 +59,6 @@ pub mod signals;
pub mod fs_ext;
pub mod history_buffer;
pub mod measured_stream;
pub mod serde_percent;