app/vminsert: returns back memory optimisation (#6794)

Production workload shows that it's useful optimisation.

Channel based objects pool allows to handle irregural data ingestion
requests and make memory allocations more smooth.
It's improves sync.Pool efficiency, since objects from sync.Pool removed
after 2 GC cycles. With GOGC=30 value, GC runs significantly more often.

https://github.com/VictoriaMetrics/VictoriaMetrics/issues/6733

### Checklist

The following checks are **mandatory**:

- [ ] My change adheres [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).

---------

Signed-off-by: f41gh7 <nik@victoriametrics.com>
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
(cherry picked from commit f255800da3)
Signed-off-by: hagen1778 <roman@victoriametrics.com>

# Conflicts:
#	app/vminsert/common/insert_ctx_pool.go
This commit is contained in:
Nikolay 2024-08-13 10:49:09 -04:00 committed by hagen1778
parent d134a310f3
commit 9f42fccfc2
No known key found for this signature in database
GPG key ID: 3BF75F3741CA9640
2 changed files with 21 additions and 5 deletions

View file

@ -2,16 +2,23 @@ package common
import (
"sync"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
)
// GetInsertCtx returns InsertCtx from the pool.
//
// Call PutInsertCtx for returning it to the pool.
func GetInsertCtx() *InsertCtx {
if v := insertCtxPool.Get(); v != nil {
return v.(*InsertCtx)
select {
case ctx := <-insertCtxPoolCh:
return ctx
default:
if v := insertCtxPool.Get(); v != nil {
return v.(*InsertCtx)
}
return &InsertCtx{}
}
return &InsertCtx{}
}
// PutInsertCtx returns ctx to the pool.
@ -19,7 +26,14 @@ func GetInsertCtx() *InsertCtx {
// ctx cannot be used after the call.
func PutInsertCtx(ctx *InsertCtx) {
ctx.Reset(0)
insertCtxPool.Put(ctx)
select {
case insertCtxPoolCh <- ctx:
default:
insertCtxPool.Put(ctx)
}
}
var insertCtxPool sync.Pool
var (
insertCtxPool sync.Pool
insertCtxPoolCh = make(chan *InsertCtx, cgroup.AvailableCPUs())
)

View file

@ -42,6 +42,8 @@ See also [LTS releases](https://docs.victoriametrics.com/lts-releases/).
* BUGFIX: [stream aggregation](https://docs.victoriametrics.com/stream-aggregation/): fix command-line flag `-remoteWrite.streamAggr.ignoreFirstIntervals` to accept multiple values and be applied per each corresponding `-remoteWrite.url`. Previously, this flag only could have been used globally for all URLs.
* BUGFIX: [graphite](https://docs.victoriametrics.com/#graphite-render-api-usage): respect `-search.denyPartialResponse` cmd-line flag and `deny_partial_response` query GET param when serving requests via Graphite API. Before, partial responses were always denied. Thanks to @penguinlav for the [pull request](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6748).
* BUGFIX: [vmagent](https://docs.victoriametrics.com/vmagent/): account for `-usePromCompatibleNaming` cmd-line flag during when pushing data to remote storages. Thanks to @12345XXX for the [pull request](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6776).
* BUGFIX: `vminsert` in [VictoriaMetrics cluster](https://docs.victoriametrics.com/cluster-victoriametrics/): reduce CPU usage by limiting the number of concurrently running inserts. The issue was introduced in [this commit](https://github.com/VictoriaMetrics/VictoriaMetrics/commit/498fe1cfa523be5bfecaa372293c3cded85e75ab) starting from v1.101.0. See [this](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/6733) issue for details.
## [v1.102.1](https://github.com/VictoriaMetrics/VictoriaMetrics/releases/tag/v1.102.1)
Released at 2024-08-01