Merge upstream prometheus/prometheus at 37b408c

.github/workflows/golangci-lint.yml

@@ -12,20 +12,21 @@ on:
 
 jobs:
   golangci:
-    name: lint
+    name: golangci-lint
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
-      - name: install Go
-        uses: actions/setup-go@v2
+        uses: actions/checkout@a5ac7e51b41094c92402da3b24376905380afc29 # v4.1.6
+      - name: Install Go
+        uses: actions/setup-go@cdcb36043654635271a94b9a6d1392de5bb323a7 # v5.0.1
         with:
-          go-version: '>=1.21 <1.22'
+          go-version: 1.22.x
       - name: Install snmp_exporter/generator dependencies
         run: sudo apt-get update && sudo apt-get -y install libsnmp-dev
         if: github.repository == 'prometheus/snmp_exporter'
       - name: Lint
-        uses: golangci/golangci-lint-action@v4.0.0
+        uses: golangci/golangci-lint-action@a4f60bb28d35aeee14e6880718e0c85ff1882e64 # v6.0.1
         with:
-          # # Make sure to sync this with Makefile.common.
-          version: v1.56.2
+          args: --verbose
+          # Make sure to sync this with Makefile.common and scripts/golangci-lint.yml.
+          version: v1.59.0

.gitpod.Dockerfile

@@ -1,15 +1,33 @@
 FROM gitpod/workspace-full
 
+# Set Node.js version as an environment variable.
 ENV CUSTOM_NODE_VERSION=16
-ENV CUSTOM_GO_VERSION=1.19
-ENV GOPATH=$HOME/go-packages
-ENV GOROOT=$HOME/go
-ENV PATH=$GOROOT/bin:$GOPATH/bin:$PATH
 
+# Install and use the specified Node.js version via nvm.
 RUN bash -c ". .nvm/nvm.sh && nvm install ${CUSTOM_NODE_VERSION} && nvm use ${CUSTOM_NODE_VERSION} && nvm alias default ${CUSTOM_NODE_VERSION}"
 
+# Ensure nvm uses the default Node.js version in all new shells.
 RUN echo "nvm use default &>/dev/null" >> ~/.bashrc.d/51-nvm-fix
-RUN curl -fsSL https://dl.google.com/go/go${GO_VERSION}.linux-amd64.tar.gz | tar xzs \
-    && printf '%s\n' 'export GOPATH=/workspace/go' \
-                      'export PATH=$GOPATH/bin:$PATH' > $HOME/.bashrc.d/300-go
 
+# Remove any existing Go installation in $HOME path.
+RUN rm -rf $HOME/go $HOME/go-packages
+
+# Export go environment variables.
+RUN echo "export GOPATH=/workspace/go" >> ~/.bashrc.d/300-go && \
+    echo "export GOBIN=\$GOPATH/bin" >> ~/.bashrc.d/300-go && \
+    echo "export GOROOT=${HOME}/go" >> ~/.bashrc.d/300-go && \
+    echo "export PATH=\$GOROOT/bin:\$GOBIN:\$PATH" >> ~/.bashrc
+
+# Reload the environment variables to ensure go environment variables are
+# available in subsequent commands.
+RUN bash -c "source ~/.bashrc && source ~/.bashrc.d/300-go"
+
+# Fetch the Go version dynamically from the Prometheus go.mod file and Install Go in $HOME path.
+RUN export CUSTOM_GO_VERSION=$(curl -sSL "https://raw.githubusercontent.com/prometheus/prometheus/main/go.mod" | awk '/^go/{print $2".0"}') && \
+    curl -fsSL "https://dl.google.com/go/go${CUSTOM_GO_VERSION}.linux-amd64.tar.gz" | \
+    tar -xz -C $HOME
+
+# Fetch the goyacc parser version dynamically from the Prometheus Makefile
+# and install it globally in $GOBIN path.
+RUN GOYACC_VERSION=$(curl -fsSL "https://raw.githubusercontent.com/prometheus/prometheus/main/Makefile" | awk -F'=' '/GOYACC_VERSION \?=/{gsub(/ /, "", $2); print $2}') && \
+    go install "golang.org/x/tools/cmd/goyacc@${GOYACC_VERSION}"

.golangci.yml

@@ -21,6 +21,7 @@ linters:
     - goimports
     - misspell
     - nolintlint
+    - perfsprint
     - predeclared
     - revive
     - testifylint
@@ -44,7 +45,9 @@ issues:
     - linters:
         - godot
       source: "^// ==="
-
+    - linters:
+        - perfsprint
+      text: "fmt.Sprintf can be replaced with string concatenation"
 linters-settings:
   depguard:
     rules:
@@ -85,6 +88,9 @@ linters-settings:
     local-prefixes: github.com/prometheus/prometheus
   gofumpt:
     extra-rules: true
+  perfsprint:
+     # Optimizes `fmt.Errorf`.
+    errorf: false
   revive:
     # By default, revive will enable only the linting rules that are named in the configuration file.
     # So, it's needed to explicitly set in configuration all required rules.

CHANGELOG.md

@@ -3,6 +3,7 @@
 ## unreleased
 
 * [CHANGE] Rules: Execute 1 query instead of N (where N is the number of alerts within alert rule) when restoring alerts. #13980
+* [FEATURE] Rules: Add new option `query_offset` for each rule group via rule group configuration file and `rule_query_offset` as part of the global configuration to have more resilience for remote write delays. #14061
 * [ENHANCEMENT] Rules: Add `rule_group_last_restore_duration_seconds` to measure the time it takes to restore a rule group. #13974
 * [ENHANCEMENT] OTLP: Improve remote write format translation performance by using label set hashes for metric identifiers instead of string based ones. #14006 #13991
 * [ENHANCEMENT] TSDB: Optimize querying with regexp matchers. #13620

CONTRIBUTING.md

@@ -42,7 +42,12 @@ go build ./cmd/prometheus/
 make test         # Make sure all the tests pass before you commit and push :)
 ```
 
-We use [`golangci-lint`](https://github.com/golangci/golangci-lint) for linting the code. If it reports an issue and you think that the warning needs to be disregarded or is a false-positive, you can add a special comment `//nolint:linter1[,linter2,...]` before the offending line. Use this sparingly though, fixing the code to comply with the linter's recommendation is in general the preferred course of action.
+To run a collection of Go linters through [`golangci-lint`](https://github.com/golangci/golangci-lint), do:
+```bash
+make lint
+```
+
+If it reports an issue and you think that the warning needs to be disregarded or is a false-positive, you can add a special comment `//nolint:linter1[,linter2,...]` before the offending line. Use this sparingly though, fixing the code to comply with the linter's recommendation is in general the preferred course of action. See [this section of the golangci-lint documentation](https://golangci-lint.run/usage/false-positives/#nolint-directive) for more information.
 
 All our issues are regularly tagged so that you can also filter down the issues involving the components you want to work on. For our labeling policy refer [the wiki page](https://github.com/prometheus/prometheus/wiki/Label-Names-and-Descriptions).
 

Makefile.common

@@ -61,7 +61,7 @@ PROMU_URL     := https://github.com/prometheus/promu/releases/download/v$(PROMU_
 SKIP_GOLANGCI_LINT :=
 GOLANGCI_LINT :=
 GOLANGCI_LINT_OPTS ?=
-GOLANGCI_LINT_VERSION ?= v1.56.2
+GOLANGCI_LINT_VERSION ?= v1.59.0
 # golangci-lint only supports linux, darwin and windows platforms on i386/amd64/arm64.
 # windows isn't included here because of the path separator being different.
 ifeq ($(GOHOSTOS),$(filter $(GOHOSTOS),linux darwin))

cmd/prometheus/main.go

@@ -42,6 +42,7 @@ import (
 	"github.com/mwitkow/go-conntrack"
 	"github.com/oklog/run"
 	"github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/client_golang/prometheus/collectors"
 	versioncollector "github.com/prometheus/client_golang/prometheus/collectors/version"
 	"github.com/prometheus/common/model"
 	"github.com/prometheus/common/promlog"
@@ -252,6 +253,18 @@ func main() {
 		newFlagRetentionDuration model.Duration
 	)
 
+	// Unregister the default GoCollector, and reregister with our defaults.
+	if prometheus.Unregister(collectors.NewGoCollector()) {
+		prometheus.MustRegister(
+			collectors.NewGoCollector(
+				collectors.WithGoCollectorRuntimeMetrics(
+					collectors.MetricsGC,
+					collectors.MetricsScheduler,
+				),
+			),
+		)
+	}
+
 	cfg := flagConfig{
 		notifier: notifier.Options{
 			Registerer: prometheus.DefaultRegisterer,
@@ -418,7 +431,7 @@ func main() {
 	serverOnlyFlag(a, "rules.alert.resend-delay", "Minimum amount of time to wait before resending an alert to Alertmanager.").
 		Default("1m").SetValue(&cfg.resendDelay)
 
-	serverOnlyFlag(a, "rules.max-concurrent-evals", "Global concurrency limit for independent rules that can run concurrently.").
+	serverOnlyFlag(a, "rules.max-concurrent-evals", "Global concurrency limit for independent rules that can run concurrently. When set, \"query.max-concurrency\" may need to be adjusted accordingly.").
 		Default("4").Int64Var(&cfg.maxConcurrentEvals)
 
 	a.Flag("scrape.adjust-timestamps", "Adjust scrape timestamps by up to `scrape.timestamp-tolerance` to align them to the intended schedule. See https://github.com/prometheus/prometheus/issues/7846 for more context. Experimental. This flag will be removed in a future release.").
@@ -772,6 +785,9 @@ func main() {
 			ResendDelay:            time.Duration(cfg.resendDelay),
 			MaxConcurrentEvals:     cfg.maxConcurrentEvals,
 			ConcurrentEvalsEnabled: cfg.enableConcurrentRuleEval,
+			DefaultRuleQueryOffset: func() time.Duration {
+				return time.Duration(cfgFile.GlobalConfig.RuleQueryOffset)
+			},
 		})
 	}
 

cmd/prometheus/main_test.go

@@ -24,6 +24,7 @@ import (
 	"os/exec"
 	"path/filepath"
 	"runtime"
+	"strconv"
 	"strings"
 	"syscall"
 	"testing"
@@ -189,7 +190,7 @@ func TestSendAlerts(t *testing.T) {
 
 	for i, tc := range testCases {
 		tc := tc
-		t.Run(fmt.Sprintf("%d", i), func(t *testing.T) {
+		t.Run(strconv.Itoa(i), func(t *testing.T) {
 			senderFunc := senderFunc(func(alerts ...*notifier.Alert) {
 				require.NotEmpty(t, tc.in, "sender called with 0 alert")
 				require.Equal(t, tc.exp, alerts)

cmd/prometheus/query_log_test.go

@@ -296,7 +296,7 @@ func (p *queryLogTest) run(t *testing.T) {
 	if p.exactQueryCount() {
 		require.Equal(t, 1, qc)
 	} else {
-		require.Greater(t, qc, 0, "no queries logged")
+		require.Positive(t, qc, "no queries logged")
 	}
 	p.validateLastQuery(t, ql)
 
@@ -366,7 +366,7 @@ func (p *queryLogTest) run(t *testing.T) {
 	if p.exactQueryCount() {
 		require.Equal(t, 1, qc)
 	} else {
-		require.Greater(t, qc, 0, "no queries logged")
+		require.Positive(t, qc, "no queries logged")
 	}
 }
 

cmd/promtool/backfill.go

@@ -88,7 +88,7 @@ func createBlocks(input []byte, mint, maxt, maxBlockDuration int64, maxSamplesIn
 	blockDuration := getCompatibleBlockDuration(maxBlockDuration)
 	mint = blockDuration * (mint / blockDuration)
 
-	db, err := tsdb.OpenDBReadOnly(outputDir, nil)
+	db, err := tsdb.OpenDBReadOnly(outputDir, "", nil)
 	if err != nil {
 		return err
 	}

cmd/promtool/main.go

@@ -235,12 +235,14 @@ func main() {
 
 	tsdbDumpCmd := tsdbCmd.Command("dump", "Dump samples from a TSDB.")
 	dumpPath := tsdbDumpCmd.Arg("db path", "Database path (default is "+defaultDBPath+").").Default(defaultDBPath).String()
+	dumpSandboxDirRoot := tsdbDumpCmd.Flag("sandbox-dir-root", "Root directory where a sandbox directory would be created in case WAL replay generates chunks. The sandbox directory is cleaned up at the end.").Default(defaultDBPath).String()
 	dumpMinTime := tsdbDumpCmd.Flag("min-time", "Minimum timestamp to dump.").Default(strconv.FormatInt(math.MinInt64, 10)).Int64()
 	dumpMaxTime := tsdbDumpCmd.Flag("max-time", "Maximum timestamp to dump.").Default(strconv.FormatInt(math.MaxInt64, 10)).Int64()
 	dumpMatch := tsdbDumpCmd.Flag("match", "Series selector. Can be specified multiple times.").Default("{__name__=~'(?s:.*)'}").Strings()
 
-	tsdbDumpOpenMetricsCmd := tsdbCmd.Command("dump-openmetrics", "[Experimental] Dump samples from a TSDB into OpenMetrics format. Native histograms are not dumped.")
+	tsdbDumpOpenMetricsCmd := tsdbCmd.Command("dump-openmetrics", "[Experimental] Dump samples from a TSDB into OpenMetrics text format, excluding native histograms and staleness markers, which are not representable in OpenMetrics.")
 	dumpOpenMetricsPath := tsdbDumpOpenMetricsCmd.Arg("db path", "Database path (default is "+defaultDBPath+").").Default(defaultDBPath).String()
+	dumpOpenMetricsSandboxDirRoot := tsdbDumpOpenMetricsCmd.Flag("sandbox-dir-root", "Root directory where a sandbox directory would be created in case WAL replay generates chunks. The sandbox directory is cleaned up at the end.").Default(defaultDBPath).String()
 	dumpOpenMetricsMinTime := tsdbDumpOpenMetricsCmd.Flag("min-time", "Minimum timestamp to dump.").Default(strconv.FormatInt(math.MinInt64, 10)).Int64()
 	dumpOpenMetricsMaxTime := tsdbDumpOpenMetricsCmd.Flag("max-time", "Maximum timestamp to dump.").Default(strconv.FormatInt(math.MaxInt64, 10)).Int64()
 	dumpOpenMetricsMatch := tsdbDumpOpenMetricsCmd.Flag("match", "Series selector. Can be specified multiple times.").Default("{__name__=~'(?s:.*)'}").Strings()
@@ -396,9 +398,9 @@ func main() {
 		os.Exit(checkErr(listBlocks(*listPath, *listHumanReadable)))
 
 	case tsdbDumpCmd.FullCommand():
-		os.Exit(checkErr(dumpSamples(ctx, *dumpPath, *dumpMinTime, *dumpMaxTime, *dumpMatch, formatSeriesSet)))
+		os.Exit(checkErr(dumpSamples(ctx, *dumpPath, *dumpSandboxDirRoot, *dumpMinTime, *dumpMaxTime, *dumpMatch, formatSeriesSet)))
 	case tsdbDumpOpenMetricsCmd.FullCommand():
-		os.Exit(checkErr(dumpSamples(ctx, *dumpOpenMetricsPath, *dumpOpenMetricsMinTime, *dumpOpenMetricsMaxTime, *dumpOpenMetricsMatch, formatSeriesSetOpenMetrics)))
+		os.Exit(checkErr(dumpSamples(ctx, *dumpOpenMetricsPath, *dumpOpenMetricsSandboxDirRoot, *dumpOpenMetricsMinTime, *dumpOpenMetricsMaxTime, *dumpOpenMetricsMatch, formatSeriesSetOpenMetrics)))
 	// TODO(aSquare14): Work on adding support for custom block size.
 	case openMetricsImportCmd.FullCommand():
 		os.Exit(backfillOpenMetrics(*importFilePath, *importDBPath, *importHumanReadable, *importQuiet, *maxBlockDuration))

cmd/promtool/main_test.go

@@ -25,6 +25,7 @@ import (
 	"os/exec"
 	"path/filepath"
 	"runtime"
+	"strconv"
 	"strings"
 	"syscall"
 	"testing"
@@ -410,7 +411,7 @@ func TestExitCodes(t *testing.T) {
 	} {
 		t.Run(c.file, func(t *testing.T) {
 			for _, lintFatal := range []bool{true, false} {
-				t.Run(fmt.Sprintf("%t", lintFatal), func(t *testing.T) {
+				t.Run(strconv.FormatBool(lintFatal), func(t *testing.T) {
 					args := []string{"-test.main", "check", "config", "testdata/" + c.file}
 					if lintFatal {
 						args = append(args, "--lint-fatal")

cmd/promtool/tsdb.go

@@ -338,7 +338,7 @@ func readPrometheusLabels(r io.Reader, n int) ([]labels.Labels, error) {
 }
 
 func listBlocks(path string, humanReadable bool) error {
-	db, err := tsdb.OpenDBReadOnly(path, nil)
+	db, err := tsdb.OpenDBReadOnly(path, "", nil)
 	if err != nil {
 		return err
 	}
@@ -393,7 +393,7 @@ func getFormatedBytes(bytes int64, humanReadable bool) string {
 }
 
 func openBlock(path, blockID string) (*tsdb.DBReadOnly, tsdb.BlockReader, error) {
-	db, err := tsdb.OpenDBReadOnly(path, nil)
+	db, err := tsdb.OpenDBReadOnly(path, "", nil)
 	if err != nil {
 		return nil, nil, err
 	}
@@ -708,8 +708,8 @@ func analyzeCompaction(ctx context.Context, block tsdb.BlockReader, indexr tsdb.
 
 type SeriesSetFormatter func(series storage.SeriesSet) error
 
-func dumpSamples(ctx context.Context, path string, mint, maxt int64, match []string, formatter SeriesSetFormatter) (err error) {
-	db, err := tsdb.OpenDBReadOnly(path, nil)
+func dumpSamples(ctx context.Context, dbDir, sandboxDirRoot string, mint, maxt int64, match []string, formatter SeriesSetFormatter) (err error) {
+	db, err := tsdb.OpenDBReadOnly(dbDir, sandboxDirRoot, nil)
 	if err != nil {
 		return err
 	}
@@ -856,9 +856,9 @@ func displayHistogram(dataType string, datas []int, total int) {
 	}
 	avg := sum / len(datas)
 	fmt.Printf("%s (min/avg/max): %d/%d/%d\n", dataType, datas[0], avg, datas[len(datas)-1])
-	maxLeftLen := strconv.Itoa(len(fmt.Sprintf("%d", end)))
-	maxRightLen := strconv.Itoa(len(fmt.Sprintf("%d", end+step)))
-	maxCountLen := strconv.Itoa(len(fmt.Sprintf("%d", maxCount)))
+	maxLeftLen := strconv.Itoa(len(strconv.Itoa(end)))
+	maxRightLen := strconv.Itoa(len(strconv.Itoa(end + step)))
+	maxCountLen := strconv.Itoa(len(strconv.Itoa(maxCount)))
 	for bucket, count := range buckets {
 		percentage := 100.0 * count / total
 		fmt.Printf("[%"+maxLeftLen+"d, %"+maxRightLen+"d]: %"+maxCountLen+"d %s\n", bucket*step+start+1, (bucket+1)*step+start, count, strings.Repeat("#", percentage))

cmd/promtool/tsdb_test.go

@@ -64,6 +64,7 @@ func getDumpedSamples(t *testing.T, path string, mint, maxt int64, match []strin
 	err := dumpSamples(
 		context.Background(),
 		path,
+		t.TempDir(),
 		mint,
 		maxt,
 		match,

cmd/promtool/unittest.go

@@ -573,7 +573,7 @@ func (la labelsAndAnnotations) String() string {
 	}
 	s := "[\n0:" + indentLines("\n"+la[0].String(), "  ")
 	for i, l := range la[1:] {
-		s += ",\n" + fmt.Sprintf("%d", i+1) + ":" + indentLines("\n"+l.String(), "  ")
+		s += ",\n" + strconv.Itoa(i+1) + ":" + indentLines("\n"+l.String(), "  ")
 	}
 	s += "\n]"
 

config/config.go

@@ -145,6 +145,7 @@ var (
 		ScrapeInterval:     model.Duration(1 * time.Minute),
 		ScrapeTimeout:      model.Duration(10 * time.Second),
 		EvaluationInterval: model.Duration(1 * time.Minute),
+		RuleQueryOffset:    model.Duration(0 * time.Minute),
 		// When native histogram feature flag is enabled, ScrapeProtocols default
 		// changes to DefaultNativeHistogramScrapeProtocols.
 		ScrapeProtocols: DefaultScrapeProtocols,
@@ -397,6 +398,8 @@ type GlobalConfig struct {
 	ScrapeProtocols []ScrapeProtocol `yaml:"scrape_protocols,omitempty"`
 	// How frequently to evaluate rules by default.
 	EvaluationInterval model.Duration `yaml:"evaluation_interval,omitempty"`
+	// Offset the rule evaluation timestamp of this particular group by the specified duration into the past to ensure the underlying metrics have been received.
+	RuleQueryOffset model.Duration `yaml:"rule_query_offset"`
 	// File to which PromQL queries are logged.
 	QueryLogFile string `yaml:"query_log_file,omitempty"`
 	// The labels to add to any timeseries that this Prometheus instance scrapes.
@@ -556,6 +559,7 @@ func (c *GlobalConfig) isZero() bool {
 		c.ScrapeInterval == 0 &&
 		c.ScrapeTimeout == 0 &&
 		c.EvaluationInterval == 0 &&
+		c.RuleQueryOffset == 0 &&
 		c.QueryLogFile == "" &&
 		c.ScrapeProtocols == nil
 }

discovery/aws/ec2.go

@@ -18,6 +18,7 @@ import (
 	"errors"
 	"fmt"
 	"net"
+	"strconv"
 	"strings"
 	"time"
 
@@ -279,7 +280,7 @@ func (d *EC2Discovery) refresh(ctx context.Context) ([]*targetgroup.Group, error
 				if inst.PrivateDnsName != nil {
 					labels[ec2LabelPrivateDNS] = model.LabelValue(*inst.PrivateDnsName)
 				}
-				addr := net.JoinHostPort(*inst.PrivateIpAddress, fmt.Sprintf("%d", d.cfg.Port))
+				addr := net.JoinHostPort(*inst.PrivateIpAddress, strconv.Itoa(d.cfg.Port))
 				labels[model.AddressLabel] = model.LabelValue(addr)
 
 				if inst.Platform != nil {

discovery/aws/lightsail.go

@@ -18,6 +18,7 @@ import (
 	"errors"
 	"fmt"
 	"net"
+	"strconv"
 	"strings"
 	"time"
 
@@ -229,7 +230,7 @@ func (d *LightsailDiscovery) refresh(ctx context.Context) ([]*targetgroup.Group,
 			lightsailLabelRegion:              model.LabelValue(d.cfg.Region),
 		}
 
-		addr := net.JoinHostPort(*inst.PrivateIpAddress, fmt.Sprintf("%d", d.cfg.Port))
+		addr := net.JoinHostPort(*inst.PrivateIpAddress, strconv.Itoa(d.cfg.Port))
 		labels[model.AddressLabel] = model.LabelValue(addr)
 
 		if inst.PublicIpAddress != nil {