rust/neon
1
0
Fork 0
mirror of https://github.com/neondatabase/neon.git synced 2026-01-05 04:22:56 +00:00
Code Issues Packages Projects Releases Wiki Activity

pageserver,safekeeper: increase heap profiling frequency to 2 MB (#10362)

Browse Source 
## Problem

Currently, the heap profiling frequency is every 1 MB allocated. Taking
a profile stack trace takes about 1 µs, and allocating 1 MB takes about
15 µs, so the overhead is about 6.7% which is a bit high. This is a
fixed cost regardless of whether heap profiles are actually accessed.

## Summary of changes

Increase the heap profiling sample frequency from 1 MB to 2 MB, which
reduces the overhead to about 3.3%. This seems acceptable, considering
performance-sensitive code will avoid allocations as far as possible
anyway.
This commit is contained in:
Erik Grinaker
2025-01-13 10:44:59 +01:00
committed by GitHub
parent 8327f68043
commit cd982a82ec
3 changed files with 10 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
pageserver/src/bin/pageserver.rs
View File

@@ -53,10 +53,12 @@ project_build_tag!(BUILD_TAG);
#[global_allocator]
static GLOBAL: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
/// Configure jemalloc to sample allocations for profiles every 1 MB (1 << 20).
/// Configure jemalloc to profile heap allocations by sampling stack traces every 2 MB (1 << 21).
/// This adds roughly 3% overhead for allocations on average, which is acceptable considering
/// performance-sensitive code will avoid allocations as far as possible anyway.
#[allow(non_upper_case_globals)]
#[export_name = "malloc_conf"]
pub static malloc_conf: &[u8] = b"prof:true,prof_active:true,lg_prof_sample:20\0";
pub static malloc_conf: &[u8] = b"prof:true,prof_active:true,lg_prof_sample:21\0";
const PID_FILE_NAME: &str = "pageserver.pid";

5
safekeeper/benches/receive_wal.rs
View File

@@ -21,14 +21,13 @@ const KB: usize = 1024;
const MB: usize = 1024 * KB;
const GB: usize = 1024 * MB;
/// Use jemalloc, and configure it to sample allocations for profiles every 1 MB.
/// This mirrors the configuration in bin/safekeeper.rs.
/// Use jemalloc and enable profiling, to mirror bin/safekeeper.rs.
#[global_allocator]
static GLOBAL: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
#[allow(non_upper_case_globals)]
#[export_name = "malloc_conf"]
pub static malloc_conf: &[u8] = b"prof:true,prof_active:true,lg_prof_sample:20\0";
pub static malloc_conf: &[u8] = b"prof:true,prof_active:true,lg_prof_sample:21\0";
// Register benchmarks with Criterion.
criterion_group!(

6
safekeeper/src/bin/safekeeper.rs
View File

@@ -51,10 +51,12 @@ use utils::{
#[global_allocator]
static GLOBAL: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
/// Configure jemalloc to sample allocations for profiles every 1 MB (1 << 20).
/// Configure jemalloc to profile heap allocations by sampling stack traces every 2 MB (1 << 21).
/// This adds roughly 3% overhead for allocations on average, which is acceptable considering
/// performance-sensitive code will avoid allocations as far as possible anyway.
#[allow(non_upper_case_globals)]
#[export_name = "malloc_conf"]
pub static malloc_conf: &[u8] = b"prof:true,prof_active:true,lg_prof_sample:20\0";
pub static malloc_conf: &[u8] = b"prof:true,prof_active:true,lg_prof_sample:21\0";
const PID_FILE_NAME: &str = "safekeeper.pid";
const ID_FILE_NAME: &str = "safekeeper.id";