all: properly handle CPU limits set on the host system/container

This can reduce memory usage on systems with enabled CPU limits.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/946

app/victoria-metrics/main.go

@@ -11,7 +11,6 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmstorage"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/buildinfo"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/envflag"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fs"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httpserver"
@@ -36,7 +35,6 @@ func main() {
 	envflag.Parse()
 	buildinfo.Init()
 	logger.Init()
-	cgroup.UpdateGOMAXPROCSToCPUQuota()
 
 	if promscrape.IsDryRun() {
 		*dryRun = true

app/vmagent/common/push_ctx.go

@@ -1,9 +1,9 @@
 package common
 
 import (
-	"runtime"
 	"sync"
 
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompbmarshal"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/promrelabel"
 )
@@ -63,4 +63,4 @@ func PutPushCtx(ctx *PushCtx) {
 }
 
 var pushCtxPool sync.Pool
-var pushCtxPoolCh = make(chan *PushCtx, runtime.GOMAXPROCS(-1))
+var pushCtxPoolCh = make(chan *PushCtx, cgroup.AvailableCPUs())

app/vmagent/influx/request_handler.go

@@ -4,12 +4,12 @@ import (
 	"flag"
 	"io"
 	"net/http"
-	"runtime"
 	"sync"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmagent/common"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmagent/remotewrite"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompbmarshal"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/promrelabel"
 	parser "github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/influx"
@@ -165,4 +165,4 @@ func putPushCtx(ctx *pushCtx) {
 }
 
 var pushCtxPool sync.Pool
-var pushCtxPoolCh = make(chan *pushCtx, runtime.GOMAXPROCS(-1))
+var pushCtxPoolCh = make(chan *pushCtx, cgroup.AvailableCPUs())

app/vmagent/main.go

@@ -21,7 +21,6 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmagent/remotewrite"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmagent/vmimport"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/buildinfo"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/envflag"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httpserver"
@@ -68,7 +67,6 @@ func main() {
 	remotewrite.InitSecretFlags()
 	buildinfo.Init()
 	logger.Init()
-	cgroup.UpdateGOMAXPROCSToCPUQuota()
 
 	if promscrape.IsDryRun() {
 		if err := promscrape.CheckConfig(); err != nil {

app/vmagent/remotewrite/remotewrite.go

@@ -3,10 +3,10 @@ package remotewrite
 import (
 	"flag"
 	"fmt"
-	"runtime"
 	"sync"
 	"sync/atomic"
 
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/decimal"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
@@ -43,7 +43,7 @@ var allRelabelConfigs atomic.Value
 
 // maxQueues limits the maximum value for `-remoteWrite.queues`. There is no sense in setting too high value,
 // since it may lead to high memory usage due to big number of buffers.
-var maxQueues = runtime.GOMAXPROCS(-1) * 4
+var maxQueues = cgroup.AvailableCPUs() * 4
 
 // InitSecretFlags must be called after flag.Parse and before any logging.
 func InitSecretFlags() {

app/vmalert/main.go

@@ -16,7 +16,6 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmalert/remoteread"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmalert/remotewrite"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/buildinfo"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/envflag"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
@@ -59,7 +58,6 @@ func main() {
 	envflag.Parse()
 	buildinfo.Init()
 	logger.Init()
-	cgroup.UpdateGOMAXPROCSToCPUQuota()
 
 	if *dryRun {
 		u, _ := url.Parse("https://victoriametrics.com/")

app/vmauth/main.go

@@ -9,7 +9,6 @@ import (
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/buildinfo"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/envflag"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httpserver"
@@ -28,7 +27,6 @@ func main() {
 	envflag.Parse()
 	buildinfo.Init()
 	logger.Init()
-	cgroup.UpdateGOMAXPROCSToCPUQuota()
 	logger.Infof("starting vmauth at %q...", *httpListenAddr)
 	startTime := time.Now()
 	initAuthConfig()

app/vmbackup/main.go

@@ -12,7 +12,6 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/backup/fslocal"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/backup/fsnil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/buildinfo"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/envflag"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
@@ -40,7 +39,6 @@ func main() {
 	envflag.Parse()
 	buildinfo.Init()
 	logger.Init()
-	cgroup.UpdateGOMAXPROCSToCPUQuota()
 
 	if len(*snapshotCreateURL) > 0 {
 		logger.Infof("Snapshots enabled")

app/vminsert/common/insert_ctx_pool.go

@@ -1,8 +1,9 @@
 package common
 
 import (
-	"runtime"
 	"sync"
+
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 )
 
 // GetInsertCtx returns InsertCtx from the pool.
@@ -33,4 +34,4 @@ func PutInsertCtx(ctx *InsertCtx) {
 }
 
 var insertCtxPool sync.Pool
-var insertCtxPoolCh = make(chan *InsertCtx, runtime.GOMAXPROCS(-1))
+var insertCtxPoolCh = make(chan *InsertCtx, cgroup.AvailableCPUs())

app/vminsert/influx/request_handler.go

@@ -4,12 +4,12 @@ import (
 	"flag"
 	"io"
 	"net/http"
-	"runtime"
 	"sync"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vminsert/common"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vminsert/relabel"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompb"
 	parser "github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/influx"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage"
@@ -175,4 +175,4 @@ func putPushCtx(ctx *pushCtx) {
 }
 
 var pushCtxPool sync.Pool
-var pushCtxPoolCh = make(chan *pushCtx, runtime.GOMAXPROCS(-1))
+var pushCtxPoolCh = make(chan *pushCtx, cgroup.AvailableCPUs())

app/vminsert/native/request_handler.go

@@ -2,11 +2,11 @@ package native
 
 import (
 	"net/http"
-	"runtime"
 	"sync"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vminsert/common"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vminsert/relabel"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompbmarshal"
 	parserCommon "github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
@@ -112,4 +112,4 @@ func putPushCtx(ctx *pushCtx) {
 }
 
 var pushCtxPool sync.Pool
-var pushCtxPoolCh = make(chan *pushCtx, runtime.GOMAXPROCS(-1))
+var pushCtxPoolCh = make(chan *pushCtx, cgroup.AvailableCPUs())

app/vminsert/vmimport/request_handler.go

@@ -2,11 +2,11 @@ package vmimport
 
 import (
 	"net/http"
-	"runtime"
 	"sync"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vminsert/common"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vminsert/relabel"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompbmarshal"
 	parserCommon "github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
@@ -117,4 +117,4 @@ func putPushCtx(ctx *pushCtx) {
 }
 
 var pushCtxPool sync.Pool
-var pushCtxPoolCh = make(chan *pushCtx, runtime.GOMAXPROCS(-1))
+var pushCtxPoolCh = make(chan *pushCtx, cgroup.AvailableCPUs())

app/vmrestore/main.go

@@ -9,7 +9,6 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/backup/common"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/backup/fslocal"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/buildinfo"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/envflag"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
@@ -33,7 +32,6 @@ func main() {
 	envflag.Parse()
 	buildinfo.Init()
 	logger.Init()
-	cgroup.UpdateGOMAXPROCSToCPUQuota()
 
 	srcFS, err := newSrcFS()
 	if err != nil {

app/vmselect/main.go

@@ -5,7 +5,6 @@ import (
 	"flag"
 	"fmt"
 	"net/http"
-	"runtime"
 	"strings"
 	"time"
 
@@ -15,6 +14,7 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/promql"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/searchutils"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmstorage"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fs"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httpserver"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
@@ -32,7 +32,7 @@ var (
 )
 
 func getDefaultMaxConcurrentRequests() int {
-	n := runtime.GOMAXPROCS(-1)
+	n := cgroup.AvailableCPUs()
 	if n <= 4 {
 		n *= 2
 	}

app/vmselect/netstorage/netstorage.go

@@ -6,7 +6,6 @@ import (
 	"flag"
 	"fmt"
 	"regexp"
-	"runtime"
 	"sort"
 	"sync"
 	"sync/atomic"
@@ -14,6 +13,7 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/searchutils"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmstorage"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage"
@@ -179,7 +179,7 @@ func (rss *Results) RunParallel(f func(rs *Result, workerID uint) error) error {
 var perQueryRowsProcessed = metrics.NewHistogram(`vm_per_query_rows_processed_count`)
 var perQuerySeriesProcessed = metrics.NewHistogram(`vm_per_query_series_processed_count`)
 
-var gomaxprocs = runtime.GOMAXPROCS(-1)
+var gomaxprocs = cgroup.AvailableCPUs()
 
 type packedTimeseries struct {
 	metricName string
@@ -265,7 +265,7 @@ func unpackWorker() {
 // unpackBatchSize is the maximum number of blocks that may be unpacked at once by a single goroutine.
 //
 // This batch is needed in order to reduce contention for upackWorkCh in multi-CPU system.
-var unpackBatchSize = 8 * runtime.GOMAXPROCS(-1)
+var unpackBatchSize = 8 * cgroup.AvailableCPUs()
 
 // Unpack unpacks pts to dst.
 func (pts *packedTimeseries) Unpack(dst *Result, tbf *tmpBlocksFile, tr storage.TimeRange, fetchData bool) error {
@@ -715,7 +715,7 @@ func ExportBlocks(sq *storage.SearchQuery, deadline searchutils.Deadline, f func
 	sr.Init(vmstorage.Storage, tfss, tr, *maxMetricsPerSearch, deadline.Deadline())
 
 	// Start workers that call f in parallel on available CPU cores.
-	gomaxprocs := runtime.GOMAXPROCS(-1)
+	gomaxprocs := cgroup.AvailableCPUs()
 	workCh := make(chan *exportWork, gomaxprocs*8)
 	var (
 		errGlobal     error

app/vmselect/prometheus/prometheus.go

@@ -5,7 +5,6 @@ import (
 	"fmt"
 	"math"
 	"net/http"
-	"runtime"
 	"sort"
 	"strconv"
 	"strings"
@@ -17,6 +16,7 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/promql"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/searchutils"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
@@ -147,7 +147,7 @@ func ExportCSVHandler(startTime time.Time, w http.ResponseWriter, r *http.Reques
 	bw := bufferedwriter.Get(w)
 	defer bufferedwriter.Put(bw)
 
-	resultsCh := make(chan *quicktemplate.ByteBuffer, runtime.GOMAXPROCS(-1))
+	resultsCh := make(chan *quicktemplate.ByteBuffer, cgroup.AvailableCPUs())
 	doneCh := make(chan error)
 	go func() {
 		err := netstorage.ExportBlocks(sq, deadline, func(mn *storage.MetricName, b *storage.Block, tr storage.TimeRange) error {
@@ -374,7 +374,7 @@ func exportHandler(w http.ResponseWriter, matches []string, start, end int64, fo
 	bw := bufferedwriter.Get(w)
 	defer bufferedwriter.Put(bw)
 
-	resultsCh := make(chan *quicktemplate.ByteBuffer, runtime.GOMAXPROCS(-1))
+	resultsCh := make(chan *quicktemplate.ByteBuffer, cgroup.AvailableCPUs())
 	doneCh := make(chan error)
 	if !reduceMemUsage {
 		rss, err := netstorage.ProcessSearchQuery(sq, true, deadline)

app/vmselect/promql/eval.go

@@ -4,12 +4,12 @@ import (
 	"flag"
 	"fmt"
 	"math"
-	"runtime"
 	"sync"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/netstorage"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/searchutils"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/memory"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage"
@@ -555,7 +555,7 @@ func evalRollupFuncWithSubquery(ec *EvalConfig, name string, rf rollupFunc, expr
 }
 
 func doParallel(tss []*timeseries, f func(ts *timeseries, values []float64, timestamps []int64) ([]float64, []int64)) {
-	concurrency := runtime.GOMAXPROCS(-1)
+	concurrency := cgroup.AvailableCPUs()
 	if concurrency > len(tss) {
 		concurrency = len(tss)
 	}
@@ -678,7 +678,7 @@ func evalRollupFuncWithMetricExpr(ec *EvalConfig, name string, rf rollupFunc,
 	timeseriesLen := rssLen
 	if iafc != nil {
 		// Incremental aggregates require holding only GOMAXPROCS timeseries in memory.
-		timeseriesLen = runtime.GOMAXPROCS(-1)
+		timeseriesLen = cgroup.AvailableCPUs()
 		if iafc.ae.Modifier.Op != "" {
 			if iafc.ae.Limit > 0 {
 				// There is an explicit limit on the number of output time series.

docs/CHANGELOG.md

@@ -9,6 +9,7 @@
 * FEATURE: export `vm_promscrape_scrapers_started_total{type="<sd_type>"}` and `vm_promscrape_scrapers_stopped_total{type="<sd_type>"}` metrics for tracking churn rate for scrapers
   per each service discovery type.
 
+* BUGFIX: properly handle CPU limits set on the host system or host container. The bugfix may result in lower memory usage on systems with CPU limits. See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/946
 * BUGFIX: prevent from duplicate `name` tag returned from `/tags/autoComplete/tags` handler. See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/942
 * BUGFIX: do not enable strict parsing for `-promscrape.config` if `-promscrape.config.dryRun` comand-line flag is set. Strict parsing can be enabled with `-promscrape.config.strictParse` command-line flag. See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/944
 

lib/cgroup/cpu.go

@@ -6,17 +6,21 @@ import (
 	"runtime"
 	"strconv"
 	"strings"
-
+	"sync"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 )
 
-// UpdateGOMAXPROCSToCPUQuota updates GOMAXPROCS to cgroup CPU quota if GOMAXPROCS isn't set in environment var.
+// AvailableCPUs returns the number of available CPU cores for the app.
-//
+func AvailableCPUs() int {
-// This function must be called after logger.Init().
+	availableCPUsOnce.Do(updateGOMAXPROCSToCPUQuota)
-func UpdateGOMAXPROCSToCPUQuota() {
+	return runtime.GOMAXPROCS(-1)
+}
+
+var availableCPUsOnce sync.Once
+
+// updateGOMAXPROCSToCPUQuota updates GOMAXPROCS to cgroup CPU quota if GOMAXPROCS isn't set in environment var.
+func updateGOMAXPROCSToCPUQuota() {
 	if v := os.Getenv("GOMAXPROCS"); v != "" {
 		// Do not override explicitly set GOMAXPROCS.
-		logger.Infof("using GOMAXPROCS=%q set via environment variable", v)
 		return
 	}
 	q := getCPUQuota()
@@ -28,13 +32,11 @@ func UpdateGOMAXPROCSToCPUQuota() {
 	numCPU := runtime.NumCPU()
 	if gomaxprocs > numCPU {
 		// There is no sense in setting more GOMAXPROCS than the number of available CPU cores.
-		logger.Infof("cgroup CPU quota=%d exceeds NumCPU=%d; using GOMAXPROCS=NumCPU", gomaxprocs, numCPU)
 		return
 	}
 	if gomaxprocs <= 0 {
 		gomaxprocs = 1
 	}
-	logger.Infof("updating GOMAXPROCS to %d according to cgroup CPU quota", gomaxprocs)
 	runtime.GOMAXPROCS(gomaxprocs)
 }
 

lib/ingestserver/graphite/server.go

@@ -4,12 +4,12 @@ import (
 	"errors"
 	"io"
 	"net"
-	"runtime"
 	"strings"
 	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/netutil"
 	"github.com/VictoriaMetrics/metrics"
@@ -113,7 +113,7 @@ func serveTCP(ln net.Listener, insertHandler func(r io.Reader) error) {
 }
 
 func serveUDP(ln net.PacketConn, insertHandler func(r io.Reader) error) {
-	gomaxprocs := runtime.GOMAXPROCS(-1)
+	gomaxprocs := cgroup.AvailableCPUs()
 	var wg sync.WaitGroup
 	for i := 0; i < gomaxprocs; i++ {
 		wg.Add(1)

lib/ingestserver/influx/server.go

@@ -4,12 +4,12 @@ import (
 	"errors"
 	"io"
 	"net"
-	"runtime"
 	"strings"
 	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/netutil"
 	"github.com/VictoriaMetrics/metrics"
@@ -113,7 +113,7 @@ func serveTCP(ln net.Listener, insertHandler func(r io.Reader) error) {
 }
 
 func serveUDP(ln net.PacketConn, insertHandler func(r io.Reader) error) {
-	gomaxprocs := runtime.GOMAXPROCS(-1)
+	gomaxprocs := cgroup.AvailableCPUs()
 	var wg sync.WaitGroup
 	for i := 0; i < gomaxprocs; i++ {
 		wg.Add(1)

lib/ingestserver/opentsdb/server.go

@@ -5,12 +5,12 @@ import (
 	"io"
 	"net"
 	"net/http"
-	"runtime"
 	"strings"
 	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/ingestserver/opentsdbhttp"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/netutil"
@@ -133,7 +133,7 @@ func serveTelnet(ln net.Listener, insertHandler func(r io.Reader) error) {
 }
 
 func serveUDP(ln net.PacketConn, insertHandler func(r io.Reader) error) {
-	gomaxprocs := runtime.GOMAXPROCS(-1)
+	gomaxprocs := cgroup.AvailableCPUs()
 	var wg sync.WaitGroup
 	for i := 0; i < gomaxprocs; i++ {
 		wg.Add(1)

lib/mergeset/table.go

@@ -6,7 +6,6 @@ import (
 	"io/ioutil"
 	"os"
 	"path/filepath"
-	"runtime"
 	"sort"
 	"strings"
 	"sync"
@@ -14,6 +13,7 @@ import (
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fs"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
@@ -959,7 +959,7 @@ func (tb *Table) maxOutPartItemsSlow() uint64 {
 	return freeSpace / uint64(mergeWorkersCount) / 4
 }
 
-var mergeWorkersCount = runtime.GOMAXPROCS(-1)
+var mergeWorkersCount = cgroup.AvailableCPUs()
 
 func openParts(path string) ([]*partWrapper, error) {
 	// The path can be missing after restoring from backup, so create it if needed.

lib/persistentqueue/fastqueue_timing_test.go

@@ -2,8 +2,9 @@ package persistentqueue
 
 import (
 	"fmt"
-	"runtime"
 	"testing"
+
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 )
 
 func BenchmarkFastQueueThroughputSerial(b *testing.B) {
@@ -36,7 +37,7 @@ func BenchmarkFastQueueThroughputConcurrent(b *testing.B) {
 			b.SetBytes(int64(blockSize) * iterationsCount)
 			path := fmt.Sprintf("bench-fast-queue-throughput-concurrent-%d", blockSize)
 			mustDeleteDir(path)
-			fq := MustOpenFastQueue(path, "foobar", iterationsCount*runtime.GOMAXPROCS(-1)*2, 0)
+			fq := MustOpenFastQueue(path, "foobar", iterationsCount*cgroup.AvailableCPUs()*2, 0)
 			defer func() {
 				fq.MustClose()
 				mustDeleteDir(path)

lib/protoparser/common/unmarshal_work.go

@@ -1,9 +1,9 @@
 package common
 
 import (
-	"runtime"
 	"sync"
 
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 )
 
@@ -25,7 +25,7 @@ func StartUnmarshalWorkers() {
 	if unmarshalWorkCh != nil {
 		logger.Panicf("BUG: it looks like startUnmarshalWorkers() has been alread called without stopUnmarshalWorkers()")
 	}
-	gomaxprocs := runtime.GOMAXPROCS(-1)
+	gomaxprocs := cgroup.AvailableCPUs()
 	unmarshalWorkCh = make(chan UnmarshalWork, 2*gomaxprocs)
 	unmarshalWorkersWG.Add(gomaxprocs)
 	for i := 0; i < gomaxprocs; i++ {

lib/protoparser/csvimport/streamparser.go

@@ -6,11 +6,11 @@ import (
 	"fmt"
 	"io"
 	"net/http"
-	"runtime"
 	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
 	"github.com/VictoriaMetrics/metrics"
 )
@@ -142,7 +142,7 @@ func putStreamContext(ctx *streamContext) {
 }
 
 var streamContextPool sync.Pool
-var streamContextPoolCh = make(chan *streamContext, runtime.GOMAXPROCS(-1))
+var streamContextPoolCh = make(chan *streamContext, cgroup.AvailableCPUs())
 
 type unmarshalWork struct {
 	rows     Rows

lib/protoparser/graphite/streamparser.go

@@ -5,11 +5,11 @@ import (
 	"flag"
 	"fmt"
 	"io"
-	"runtime"
 	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
 	"github.com/VictoriaMetrics/metrics"
@@ -127,7 +127,7 @@ func putStreamContext(ctx *streamContext) {
 }
 
 var streamContextPool sync.Pool
-var streamContextPoolCh = make(chan *streamContext, runtime.GOMAXPROCS(-1))
+var streamContextPoolCh = make(chan *streamContext, cgroup.AvailableCPUs())
 
 type unmarshalWork struct {
 	rows     Rows

lib/protoparser/influx/streamparser.go

@@ -5,11 +5,11 @@ import (
 	"flag"
 	"fmt"
 	"io"
-	"runtime"
 	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
 	"github.com/VictoriaMetrics/metrics"
@@ -155,7 +155,7 @@ func putStreamContext(ctx *streamContext) {
 }
 
 var streamContextPool sync.Pool
-var streamContextPoolCh = make(chan *streamContext, runtime.GOMAXPROCS(-1))
+var streamContextPoolCh = make(chan *streamContext, cgroup.AvailableCPUs())
 
 type unmarshalWork struct {
 	rows         Rows

lib/protoparser/opentsdb/streamparser.go

@@ -5,11 +5,11 @@ import (
 	"flag"
 	"fmt"
 	"io"
-	"runtime"
 	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
 	"github.com/VictoriaMetrics/metrics"
@@ -126,7 +126,7 @@ func putStreamContext(ctx *streamContext) {
 }
 
 var streamContextPool sync.Pool
-var streamContextPoolCh = make(chan *streamContext, runtime.GOMAXPROCS(-1))
+var streamContextPoolCh = make(chan *streamContext, cgroup.AvailableCPUs())
 
 type unmarshalWork struct {
 	rows     Rows

lib/protoparser/opentsdbhttp/streamparser.go

@@ -6,11 +6,11 @@ import (
 	"fmt"
 	"io"
 	"net/http"
-	"runtime"
 	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
@@ -149,7 +149,7 @@ func putStreamContext(ctx *streamContext) {
 }
 
 var streamContextPool sync.Pool
-var streamContextPoolCh = make(chan *streamContext, runtime.GOMAXPROCS(-1))
+var streamContextPoolCh = make(chan *streamContext, cgroup.AvailableCPUs())
 
 func getRows() *Rows {
 	v := rowsPool.Get()

lib/protoparser/prometheus/streamparser.go

@@ -4,11 +4,11 @@ import (
 	"bufio"
 	"fmt"
 	"io"
-	"runtime"
 	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
 	"github.com/VictoriaMetrics/metrics"
 )
@@ -128,7 +128,7 @@ func putStreamContext(ctx *streamContext) {
 }
 
 var streamContextPool sync.Pool
-var streamContextPoolCh = make(chan *streamContext, runtime.GOMAXPROCS(-1))
+var streamContextPoolCh = make(chan *streamContext, cgroup.AvailableCPUs())
 
 type unmarshalWork struct {
 	rows             Rows

lib/protoparser/promremotewrite/streamparser.go

@@ -5,10 +5,10 @@ import (
 	"fmt"
 	"io"
 	"net/http"
-	"runtime"
 	"sync"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompb"
@@ -123,7 +123,7 @@ func putPushCtx(ctx *pushCtx) {
 }
 
 var pushCtxPool sync.Pool
-var pushCtxPoolCh = make(chan *pushCtx, runtime.GOMAXPROCS(-1))
+var pushCtxPoolCh = make(chan *pushCtx, cgroup.AvailableCPUs())
 
 func getWriteRequest() *prompb.WriteRequest {
 	v := writeRequestPool.Get()

lib/protoparser/vmimport/streamparser.go

@@ -5,10 +5,10 @@ import (
 	"fmt"
 	"io"
 	"net/http"
-	"runtime"
 	"sync"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
 	"github.com/VictoriaMetrics/metrics"
@@ -132,7 +132,7 @@ func putStreamContext(ctx *streamContext) {
 }
 
 var streamContextPool sync.Pool
-var streamContextPoolCh = make(chan *streamContext, runtime.GOMAXPROCS(-1))
+var streamContextPoolCh = make(chan *streamContext, cgroup.AvailableCPUs())
 
 type unmarshalWork struct {
 	rows     Rows

lib/storage/partition.go

@@ -7,7 +7,6 @@ import (
 	"math/bits"
 	"os"
 	"path/filepath"
-	"runtime"
 	"sort"
 	"strings"
 	"sync"
@@ -16,6 +15,7 @@ import (
 	"unsafe"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fs"
@@ -77,7 +77,7 @@ const finalPartsToMerge = 3
 // The number of shards for rawRow entries per partition.
 //
 // Higher number of shards reduces CPU contention and increases the max bandwidth on multi-core systems.
-var rawRowsShardsPerPartition = (runtime.GOMAXPROCS(-1) + 7) / 8
+var rawRowsShardsPerPartition = (cgroup.AvailableCPUs() + 7) / 8
 
 // getMaxRowsPerPartition returns the maximum number of rows that haven't been converted into parts yet.
 func getMaxRawRowsPerPartition() int {
@@ -882,8 +882,8 @@ func hasActiveMerges(pws []*partWrapper) bool {
 }
 
 var (
-	bigMergeWorkersCount   = (runtime.GOMAXPROCS(-1) + 1) / 2
+	bigMergeWorkersCount   = (cgroup.AvailableCPUs() + 1) / 2
-	smallMergeWorkersCount = (runtime.GOMAXPROCS(-1) + 1) / 2
+	smallMergeWorkersCount = (cgroup.AvailableCPUs() + 1) / 2
 )
 
 // SetBigMergeWorkersCount sets the maximum number of concurrent mergers for big blocks.

lib/storage/storage.go

@@ -8,13 +8,13 @@ import (
 	"os"
 	"path/filepath"
 	"regexp"
-	"runtime"
 	"sort"
 	"sync"
 	"sync/atomic"
 	"time"
 	"unsafe"
 
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fs"
@@ -873,7 +873,7 @@ var (
 	// Limit the concurrency for TSID searches to GOMAXPROCS*2, since this operation
 	// is CPU bound and sometimes disk IO bound, so there is no sense in running more
 	// than GOMAXPROCS*2 concurrent goroutines for TSID searches.
-	searchTSIDsConcurrencyCh = make(chan struct{}, runtime.GOMAXPROCS(-1)*2)
+	searchTSIDsConcurrencyCh = make(chan struct{}, cgroup.AvailableCPUs()*2)
 )
 
 // prefetchMetricNames pre-fetches metric names for the given tsids into metricID->metricName cache.
@@ -1149,7 +1149,7 @@ var (
 	// Limit the concurrency for data ingestion to GOMAXPROCS, since this operation
 	// is CPU bound, so there is no sense in running more than GOMAXPROCS concurrent
 	// goroutines on data ingestion path.
-	addRowsConcurrencyCh = make(chan struct{}, runtime.GOMAXPROCS(-1))
+	addRowsConcurrencyCh = make(chan struct{}, cgroup.AvailableCPUs())
 	addRowsTimeout       = 30 * time.Second
 )
 

lib/storage/table_search_timing_test.go

@@ -4,11 +4,12 @@ import (
 	"fmt"
 	"math/rand"
 	"os"
-	"runtime"
 	"sync"
 	"sync/atomic"
 	"testing"
 	"time"
+
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 )
 
 func TestMain(m *testing.M) {
@@ -79,7 +80,7 @@ func createBenchTable(b *testing.B, path string, startTimestamp int64, rowsPerIn
 	timestamp := uint64(startTimestamp)
 
 	var wg sync.WaitGroup
-	for k := 0; k < runtime.GOMAXPROCS(-1); k++ {
+	for k := 0; k < cgroup.AvailableCPUs(); k++ {
 		wg.Add(1)
 		go func() {
 			rows := make([]rawRow, rowsPerInsert)

lib/storage/table_timing_test.go

@@ -4,9 +4,10 @@ import (
 	"fmt"
 	"math/rand"
 	"os"
-	"runtime"
 	"testing"
 	"time"
+
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 )
 
 func BenchmarkTableAddRows(b *testing.B) {
@@ -57,7 +58,7 @@ func benchmarkTableAddRows(b *testing.B, rowsPerInsert, tsidsCount int) {
 		close(workCh)
 
 		doneCh := make(chan struct{})
-		gomaxprocs := runtime.GOMAXPROCS(-1)
+		gomaxprocs := cgroup.AvailableCPUs()
 
 		for j := 0; j < gomaxprocs; j++ {
 			go func(goroutineID int) {

lib/workingsetcache/cache.go

@@ -1,11 +1,11 @@
 package workingsetcache
 
 import (
-	"runtime"
 	"sync"
 	"sync/atomic"
 	"time"
 
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/fastcache"
 )
 
@@ -198,7 +198,7 @@ func (c *Cache) cacheSizeWatcher(maxBytes int) {
 // Save safes the cache to filePath.
 func (c *Cache) Save(filePath string) error {
 	curr := c.curr.Load().(*fastcache.Cache)
-	concurrency := runtime.GOMAXPROCS(-1)
+	concurrency := cgroup.AvailableCPUs()
 	return curr.SaveToFileConcurrent(filePath, concurrency)
 }
 

lib/writeconcurrencylimiter/concurrencylimiter.go

@@ -4,16 +4,16 @@ import (
 	"flag"
 	"fmt"
 	"net/http"
-	"runtime"
 	"time"
 
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httpserver"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/timerpool"
 	"github.com/VictoriaMetrics/metrics"
 )
 
 var (
-	maxConcurrentInserts = flag.Int("maxConcurrentInserts", runtime.GOMAXPROCS(-1)*4, "The maximum number of concurrent inserts. Default value should work for most cases, "+
+	maxConcurrentInserts = flag.Int("maxConcurrentInserts", cgroup.AvailableCPUs()*4, "The maximum number of concurrent inserts. Default value should work for most cases, "+
 		"since it minimizes the overhead for concurrent inserts. This option is tigthly coupled with -insert.maxQueueDuration")
 	maxQueueDuration = flag.Duration("insert.maxQueueDuration", time.Minute, "The maximum duration for waiting in the queue for insert requests due to -maxConcurrentInserts")
 )