package main
import (
"embed"
"errors"
"flag"
"fmt"
"net/http"
"net/http/httputil"
"net/url"
"os"
"strings"
"time"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/clusternative"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/graphite"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/netstorage"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/prometheus"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/promql"
"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/searchutils"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/auth"
"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/fs"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/httpserver"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/httputils"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/procutil"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/promrelabel"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/pushmetrics"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/querytracer"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/tenantmetrics"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/timerpool"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/vmselectapi"
"github.com/VictoriaMetrics/metrics"
)
var (
httpListenAddrs = flagutil.NewArrayString("httpListenAddr", "Address to listen for incoming http requests. See also -httpListenAddr.useProxyProtocol")
useProxyProtocol = flagutil.NewArrayBool("httpListenAddr.useProxyProtocol", "Whether to use proxy protocol for connections accepted at the given -httpListenAddr . "+
"See https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt . "+
"With enabled proxy protocol http server cannot serve regular /metrics endpoint. Use -pushmetrics.url for metrics pushing")
cacheDataPath = flag.String("cacheDataPath", "", "Path to directory for cache files. Cache isn't saved if empty")
maxConcurrentRequests = flag.Int("search.maxConcurrentRequests", getDefaultMaxConcurrentRequests(), "The maximum number of concurrent search requests. "+
"It shouldn't be high, since a single request can saturate all the CPU cores, while many concurrently executed requests may require high amounts of memory. "+
"See also -search.maxQueueDuration and -search.maxMemoryPerQuery")
maxQueueDuration = flag.Duration("search.maxQueueDuration", 10*time.Second, "The maximum time the request waits for execution when -search.maxConcurrentRequests "+
"limit is reached; see also -search.maxQueryDuration")
minScrapeInterval = flag.Duration("dedup.minScrapeInterval", 0, "Leave only the last sample in every time series per each discrete interval "+
"equal to -dedup.minScrapeInterval > 0. See https://docs.victoriametrics.com/#deduplication for details")
deleteAuthKey = flagutil.NewPassword("deleteAuthKey", "authKey for metrics' deletion via /prometheus/api/v1/admin/tsdb/delete_series and /graphite/tags/delSeries")
resetCacheAuthKey = flagutil.NewPassword("search.resetCacheAuthKey", "Optional authKey for resetting rollup cache via /internal/resetRollupResultCache call")
logSlowQueryDuration = flag.Duration("search.logSlowQueryDuration", 5*time.Second, "Log queries with execution time exceeding this value. Zero disables slow query logging. "+
"See also -search.logQueryMemoryUsage")
vmalertProxyURL = flag.String("vmalert.proxyURL", "", "Optional URL for proxying requests to vmalert. For example, if -vmalert.proxyURL=http://vmalert:8880 , then alerting API requests such as /api/v1/rules from Grafana will be proxied to http://vmalert:8880/api/v1/rules")
storageNodes = flagutil.NewArrayString("storageNode", "Comma-separated addresses of vmstorage nodes; usage: -storageNode=vmstorage-host1,...,vmstorage-hostN . "+
"Enterprise version of VictoriaMetrics supports automatic discovery of vmstorage addresses via DNS SRV records. For example, -storageNode=srv+vmstorage.addrs . "+
"See https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html#automatic-vmstorage-discovery")
clusternativeListenAddr = flag.String("clusternativeListenAddr", "", "TCP address to listen for requests from other vmselect nodes in multi-level cluster setup. "+
"See https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html#multi-level-cluster-setup . Usually :8401 should be set to match default vmstorage port for vmselect. Disabled work if empty")
)
var slowQueries = metrics.NewCounter(`vm_slow_queries_total`)
func getDefaultMaxConcurrentRequests() int {
n := cgroup.AvailableCPUs()
if n <= 4 {
n *= 2
}
if n > 16 {
// A single request can saturate all the CPU cores, so there is no sense
// in allowing higher number of concurrent requests - they will just contend
// for unavailable CPU time.
n = 16
}
return n
}
//go:embed static
var staticFiles embed.FS
var staticServer = http.FileServer(http.FS(staticFiles))
func main() {
// Write flags and help message to stdout, since it is easier to grep or pipe.
flag.CommandLine.SetOutput(os.Stdout)
flag.Usage = usage
envflag.Parse()
buildinfo.Init()
logger.Init()
logger.Infof("starting netstorage at storageNodes %s", *storageNodes)
startTime := time.Now()
storage.SetDedupInterval(*minScrapeInterval)
if len(*storageNodes) == 0 {
logger.Fatalf("missing -storageNode arg")
}
if hasEmptyValues(*storageNodes) {
logger.Fatalf("found empty address of storage node in the -storageNodes flag, please make sure that all -storageNode args are non-empty")
}
if duplicatedAddr := checkDuplicates(*storageNodes); duplicatedAddr != "" {
logger.Fatalf("found equal addresses of storage nodes in the -storageNodes flag: %q", duplicatedAddr)
}
netstorage.Init(*storageNodes)
logger.Infof("started netstorage in %.3f seconds", time.Since(startTime).Seconds())
if len(*cacheDataPath) > 0 {
tmpDataPath := *cacheDataPath + "/tmp"
fs.RemoveDirContents(tmpDataPath)
netstorage.InitTmpBlocksDir(tmpDataPath)
promql.InitRollupResultCache(*cacheDataPath + "/rollupResult")
} else {
netstorage.InitTmpBlocksDir("")
promql.InitRollupResultCache("")
}
concurrencyLimitCh = make(chan struct{}, *maxConcurrentRequests)
initVMAlertProxy()
var vmselectapiServer *vmselectapi.Server
if *clusternativeListenAddr != "" {
logger.Infof("starting vmselectapi server at %q", *clusternativeListenAddr)
s, err := clusternative.NewVMSelectServer(*clusternativeListenAddr)
if err != nil {
logger.Fatalf("cannot initialize vmselectapi server: %s", err)
}
vmselectapiServer = s
logger.Infof("started vmselectapi server at %q", *clusternativeListenAddr)
}
listenAddrs := *httpListenAddrs
if len(listenAddrs) == 0 {
listenAddrs = []string{":8481"}
}
go httpserver.Serve(listenAddrs, useProxyProtocol, requestHandler)
pushmetrics.Init()
sig := procutil.WaitForSigterm()
logger.Infof("service received signal %s", sig)
pushmetrics.Stop()
logger.Infof("gracefully shutting down http service at %q", listenAddrs)
startTime = time.Now()
if err := httpserver.Stop(listenAddrs); err != nil {
logger.Fatalf("cannot stop http service: %s", err)
}
logger.Infof("successfully shut down http service in %.3f seconds", time.Since(startTime).Seconds())
if vmselectapiServer != nil {
logger.Infof("stopping vmselectapi server...")
vmselectapiServer.MustStop()
logger.Infof("stopped vmselectapi server")
}
logger.Infof("shutting down neststorage...")
startTime = time.Now()
netstorage.MustStop()
if len(*cacheDataPath) > 0 {
promql.StopRollupResultCache()
}
logger.Infof("successfully stopped netstorage in %.3f seconds", time.Since(startTime).Seconds())
fs.MustStopDirRemover()
logger.Infof("the vmselect has been stopped")
}
var concurrencyLimitCh chan struct{}
var (
concurrencyLimitReached = metrics.NewCounter(`vm_concurrent_select_limit_reached_total`)
concurrencyLimitTimeout = metrics.NewCounter(`vm_concurrent_select_limit_timeout_total`)
_ = metrics.NewGauge(`vm_concurrent_select_capacity`, func() float64 {
return float64(cap(concurrencyLimitCh))
})
_ = metrics.NewGauge(`vm_concurrent_select_current`, func() float64 {
return float64(len(concurrencyLimitCh))
})
)
func requestHandler(w http.ResponseWriter, r *http.Request) bool {
path := strings.Replace(r.URL.Path, "//", "/", -1)
if handleStaticAndSimpleRequests(w, r, path) {
return true
}
// Handle non-trivial dynamic requests, which may take big amounts of time and resources.
startTime := time.Now()
defer requestDuration.UpdateDuration(startTime)
tracerEnabled := httputils.GetBool(r, "trace")
qt := querytracer.New(tracerEnabled, r.URL.Path)
// Limit the number of concurrent queries.
select {
case concurrencyLimitCh <- struct{}{}:
defer func() { <-concurrencyLimitCh }()
default:
// Sleep for a while until giving up. This should resolve short bursts in requests.
concurrencyLimitReached.Inc()
d := searchutils.GetMaxQueryDuration(r)
if d > *maxQueueDuration {
d = *maxQueueDuration
}
t := timerpool.Get(d)
select {
case concurrencyLimitCh <- struct{}{}:
timerpool.Put(t)
qt.Printf("wait in queue because -search.maxConcurrentRequests=%d concurrent requests are executed", *maxConcurrentRequests)
defer func() { <-concurrencyLimitCh }()
case <-r.Context().Done():
timerpool.Put(t)
remoteAddr := httpserver.GetQuotedRemoteAddr(r)
requestURI := httpserver.GetRequestURI(r)
logger.Infof("client has cancelled the request after %.3f seconds: remoteAddr=%s, requestURI: %q",
time.Since(startTime).Seconds(), remoteAddr, requestURI)
return true
case <-t.C:
timerpool.Put(t)
concurrencyLimitTimeout.Inc()
err := &httpserver.ErrorWithStatusCode{
Err: fmt.Errorf("couldn't start executing the request in %.3f seconds, since -search.maxConcurrentRequests=%d concurrent requests "+
"are executed. Possible solutions: to reduce query load; to add more compute resources to the server; "+
"to increase -search.maxQueueDuration=%s; to increase -search.maxQueryDuration; to increase -search.maxConcurrentRequests",
d.Seconds(), *maxConcurrentRequests, maxQueueDuration),
StatusCode: http.StatusTooManyRequests,
}
w.Header().Add("Retry-After", "10")
httpserver.Errorf(w, r, "%s", err)
return true
}
}
if *logSlowQueryDuration > 0 {
actualStartTime := time.Now()
defer func() {
d := time.Since(actualStartTime)
if d >= *logSlowQueryDuration {
remoteAddr := httpserver.GetQuotedRemoteAddr(r)
requestURI := httpserver.GetRequestURI(r)
logger.Warnf("slow query according to -search.logSlowQueryDuration=%s: remoteAddr=%s, duration=%.3f seconds; requestURI: %q",
*logSlowQueryDuration, remoteAddr, d.Seconds(), requestURI)
slowQueries.Inc()
}
}()
}
if path == "/internal/resetRollupResultCache" {
if !httpserver.CheckAuthFlag(w, r, resetCacheAuthKey.Get(), "resetCacheAuthKey") {
return true
}
promql.ResetRollupResultCache()
return true
}
if path == "/admin/tenants" {
tenantsRequests.Inc()
httpserver.EnableCORS(w, r)
if err := prometheus.Tenants(qt, startTime, w, r); err != nil {
tenantsErrors.Inc()
httpserver.Errorf(w, r, "error getting tenants: %s", err)
return true
}
return true
}
p, err := httpserver.ParsePath(path)
if err != nil {
httpserver.Errorf(w, r, "cannot parse path %q: %s", path, err)
return true
}
at, err := auth.NewToken(p.AuthToken)
if err != nil {
httpserver.Errorf(w, r, "auth error: %s", err)
return true
}
switch p.Prefix {
case "select":
return selectHandler(qt, startTime, w, r, p, at)
case "delete":
return deleteHandler(startTime, w, r, p, at)
default:
// This is not our link
return false
}
}
//go:embed vmui
var vmuiFiles embed.FS
var vmuiFileServer = http.FileServer(http.FS(vmuiFiles))
func selectHandler(qt *querytracer.Tracer, startTime time.Time, w http.ResponseWriter, r *http.Request, p *httpserver.Path, at *auth.Token) bool {
defer func() {
// Count per-tenant cumulative durations and total requests
httpRequests.Get(at).Inc()
httpRequestsDuration.Get(at).Add(int(time.Since(startTime).Milliseconds()))
}()
if strings.HasPrefix(p.Suffix, "prometheus/api/v1/label/") {
s := p.Suffix[len("prometheus/api/v1/label/"):]
if strings.HasSuffix(s, "/values") {
labelValuesRequests.Inc()
labelName := s[:len(s)-len("/values")]
httpserver.EnableCORS(w, r)
if err := prometheus.LabelValuesHandler(qt, startTime, at, labelName, w, r); err != nil {
labelValuesErrors.Inc()
sendPrometheusError(w, r, err)
return true
}
return true
}
}
if strings.HasPrefix(p.Suffix, "graphite/tags/") && !isGraphiteTagsPath(p.Suffix[len("graphite"):]) {
tagName := p.Suffix[len("graphite/tags/"):]
graphiteTagValuesRequests.Inc()
if err := graphite.TagValuesHandler(startTime, at, tagName, w, r); err != nil {
graphiteTagValuesErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
}
switch p.Suffix {
case "prometheus/api/v1/query":
queryRequests.Inc()
httpserver.EnableCORS(w, r)
if err := prometheus.QueryHandler(qt, startTime, at, w, r); err != nil {
queryErrors.Inc()
sendPrometheusError(w, r, err)
return true
}
return true
case "prometheus/api/v1/query_range":
queryRangeRequests.Inc()
httpserver.EnableCORS(w, r)
if err := prometheus.QueryRangeHandler(qt, startTime, at, w, r); err != nil {
queryRangeErrors.Inc()
sendPrometheusError(w, r, err)
return true
}
return true
case "prometheus/api/v1/series":
seriesRequests.Inc()
httpserver.EnableCORS(w, r)
if err := prometheus.SeriesHandler(qt, startTime, at, w, r); err != nil {
seriesErrors.Inc()
sendPrometheusError(w, r, err)
return true
}
return true
case "prometheus/api/v1/series/count":
seriesCountRequests.Inc()
httpserver.EnableCORS(w, r)
if err := prometheus.SeriesCountHandler(startTime, at, w, r); err != nil {
seriesCountErrors.Inc()
sendPrometheusError(w, r, err)
return true
}
return true
case "prometheus/api/v1/labels":
labelsRequests.Inc()
httpserver.EnableCORS(w, r)
if err := prometheus.LabelsHandler(qt, startTime, at, w, r); err != nil {
labelsErrors.Inc()
sendPrometheusError(w, r, err)
return true
}
return true
case "prometheus/api/v1/status/tsdb":
statusTSDBRequests.Inc()
httpserver.EnableCORS(w, r)
if err := prometheus.TSDBStatusHandler(qt, startTime, at, w, r); err != nil {
statusTSDBErrors.Inc()
sendPrometheusError(w, r, err)
return true
}
return true
case "prometheus/api/v1/export":
exportRequests.Inc()
if err := prometheus.ExportHandler(startTime, at, w, r); err != nil {
exportErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "prometheus/api/v1/export/csv":
exportCSVRequests.Inc()
if err := prometheus.ExportCSVHandler(startTime, at, w, r); err != nil {
exportCSVErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "prometheus/api/v1/export/native":
exportNativeRequests.Inc()
if err := prometheus.ExportNativeHandler(startTime, at, w, r); err != nil {
exportNativeErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "prometheus/federate":
federateRequests.Inc()
if err := prometheus.FederateHandler(startTime, at, w, r); err != nil {
federateErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/metrics/find", "graphite/metrics/find/":
graphiteMetricsFindRequests.Inc()
httpserver.EnableCORS(w, r)
if err := graphite.MetricsFindHandler(startTime, at, w, r); err != nil {
graphiteMetricsFindErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/metrics/expand", "graphite/metrics/expand/":
graphiteMetricsExpandRequests.Inc()
httpserver.EnableCORS(w, r)
if err := graphite.MetricsExpandHandler(startTime, at, w, r); err != nil {
graphiteMetricsExpandErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/metrics/index.json", "graphite/metrics/index.json/":
graphiteMetricsIndexRequests.Inc()
httpserver.EnableCORS(w, r)
if err := graphite.MetricsIndexHandler(startTime, at, w, r); err != nil {
graphiteMetricsIndexErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/tags/tagSeries":
graphiteTagsTagSeriesRequests.Inc()
if err := graphite.TagsTagSeriesHandler(startTime, at, w, r); err != nil {
graphiteTagsTagSeriesErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/tags/tagMultiSeries":
graphiteTagsTagMultiSeriesRequests.Inc()
if err := graphite.TagsTagMultiSeriesHandler(startTime, at, w, r); err != nil {
graphiteTagsTagMultiSeriesErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/tags":
graphiteTagsRequests.Inc()
if err := graphite.TagsHandler(startTime, at, w, r); err != nil {
graphiteTagsErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/tags/findSeries":
graphiteTagsFindSeriesRequests.Inc()
if err := graphite.TagsFindSeriesHandler(startTime, at, w, r); err != nil {
graphiteTagsFindSeriesErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/tags/autoComplete/tags":
graphiteTagsAutoCompleteTagsRequests.Inc()
httpserver.EnableCORS(w, r)
if err := graphite.TagsAutoCompleteTagsHandler(startTime, at, w, r); err != nil {
graphiteTagsAutoCompleteTagsErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/tags/autoComplete/values":
graphiteTagsAutoCompleteValuesRequests.Inc()
httpserver.EnableCORS(w, r)
if err := graphite.TagsAutoCompleteValuesHandler(startTime, at, w, r); err != nil {
graphiteTagsAutoCompleteValuesErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/tags/delSeries":
if !httpserver.CheckAuthFlag(w, r, deleteAuthKey.Get(), "deleteAuthKey") {
return true
}
graphiteTagsDelSeriesRequests.Inc()
if err := graphite.TagsDelSeriesHandler(startTime, at, w, r); err != nil {
graphiteTagsDelSeriesErrors.Inc()
httpserver.Errorf(w, r, "%s", err)
return true
}
return true
case "graphite/render":
graphiteRenderRequests.Inc()
if err := graphite.RenderHandler(startTime, at, w, r); err != nil {
graphiteRenderErrors.Inc()
httpserver.Errorf(w, r, "error in %q: %s", r.URL.Path, err)
return true
}
return true
default:
return false
}
}
func handleStaticAndSimpleRequests(w http.ResponseWriter, r *http.Request, path string) bool {
if path == "/" {
if r.Method != http.MethodGet {
return false
}
w.Header().Add("Content-Type", "text/html; charset=utf-8")
fmt.Fprintf(w, `vmselect - a component of VictoriaMetrics cluster
docs
`)
return true
}
if path == "/api/v1/status/top_queries" {
globalTopQueriesRequests.Inc()
httpserver.EnableCORS(w, r)
if err := prometheus.QueryStatsHandler(nil, w, r); err != nil {
globalTopQueriesErrors.Inc()
sendPrometheusError(w, r, err)
return true
}
return true
}
if path == "/api/v1/status/active_queries" {
globalStatusActiveQueriesRequests.Inc()
httpserver.EnableCORS(w, r)
promql.ActiveQueriesHandler(nil, w, r)
return true
}
p, err := httpserver.ParsePath(path)
if err != nil {
return false
}
if p.Suffix == "" {
if r.Method != http.MethodGet {
return false
}
w.Header().Add("Content-Type", "text/html; charset=utf-8")
fmt.Fprintf(w, "