Merge branch 'master' into dev-2.0

.travis.yml

@@ -3,11 +3,9 @@ sudo: false
 language: go
 
 go:
-- 1.8
+- 1.8.x
 
 go_import_path: github.com/prometheus/prometheus
 
 script:
 - make check_license style test
-
-

CHANGELOG.md

@@ -1,3 +1,72 @@
+## 1.6.1 / 2017-04-19
+
+* [BUGFIX] Don't panic if storage has no FPs even after initial wait
+
+## 1.6.0 / 2017-04-14
+
+* [CHANGE] Replaced the remote write implementations for various backends by a
+  generic write interface with example adapter implementation for various
+  backends. Note that both the previous and the current remote write
+  implementations are **experimental**.
+* [FEATURE] New flag `-storage.local.target-heap-size` to tell Prometheus about
+  the desired heap size. This deprecates the flags
+  `-storage.local.memory-chunks` and `-storage.local.max-chunks-to-persist`,
+  which are kept for backward compatibility.
+* [FEATURE] Add `check-metrics` to `promtool` to lint metric names.
+* [FEATURE] Add Joyent Triton discovery.
+* [FEATURE] `X-Prometheus-Scrape-Timeout-Seconds` header in HTTP scrape
+  requests.
+* [FEATURE] Remote read interface, including example for InfluxDB. **Experimental.**
+* [FEATURE] Enable Consul SD to connect via TLS.
+* [FEATURE] Marathon SD supports multiple ports.
+* [FEATURE] Marathon SD supports bearer token for authentication.
+* [FEATURE] Custom timeout for queries.
+* [FEATURE] Expose `buildQueryUrl` in `graph.js`.
+* [FEATURE] Add `rickshawGraph` property to the graph object in console
+  templates.
+* [FEATURE] New metrics exported by Prometheus itself:
+  * Summary `prometheus_engine_query_duration_seconds`
+  * Counter `prometheus_evaluator_iterations_missed_total`
+  * Counter `prometheus_evaluator_iterations_total`
+  * Gauge `prometheus_local_storage_open_head_chunks`
+  * Gauge `prometheus_local_storage_target_heap_size`
+* [ENHANCEMENT] Reduce shut-down time by interrupting an ongoing checkpoint
+  before starting the final checkpoint.
+* [ENHANCEMENT] Auto-tweak times between checkpoints to limit time spent in
+  checkpointing to 50%.
+* [ENHANCEMENT] Improved crash recovery deals better with certain index
+  corruptions.
+* [ENHANCEMENT] Graphing deals better with constant time series.
+* [ENHANCEMENT] Retry remote writes on recoverable errors.
+* [ENHANCEMENT] Evict unused chunk descriptors during crash recovery to limit
+  memory usage.
+* [ENHANCEMENT] Smoother disk usage during series maintenance.
+* [ENHANCEMENT] Targets on targets page sorted by instance within a job.
+* [ENHANCEMENT] Sort labels in federation.
+* [ENHANCEMENT] Set `GOGC=40` by default, which results in much better memory
+  utilization at the price of slightly higher CPU usage. If `GOGC` is set by
+  the user, it is still honored as usual.
+* [ENHANCEMENT] Close head chunks after being idle for the duration of the
+  configured staleness delta. This helps to persist and evict head chunk of
+  stale series more quickly.
+* [ENHANCEMENT] Stricter checking of relabel config.
+* [ENHANCEMENT] Cache busters for static web content.
+* [ENHANCEMENT] Send Prometheus-specific user-agent header during scrapes.
+* [ENHANCEMENT] Improved performance of series retention cut-off.
+* [ENHANCEMENT] Mitigate impact of non-atomic sample ingestion on
+  `histogram_quantile` by enforcing buckets to be monotonic.
+* [ENHANCEMENT] Released binaries built with Go 1.8.1.
+* [BUGFIX] Send `instance=""` with federation if `instance` not set.
+* [BUGFIX] Update to new `client_golang` to get rid of unwanted quantile
+  metrics in summaries.
+* [BUGFIX] Introduce several additional guards against data corruption.
+* [BUGFIX] Mark storage dirty and increment
+  `prometheus_local_storage_persist_errors_total` on all relevant errors.
+* [BUGFIX] Propagate storage errors as 500 in the HTTP API.
+* [BUGFIX] Fix int64 overflow in timestamps in the HTTP API.
+* [BUGFIX] Fix deadlock in Zookeeper SD.
+* [BUGFIX] Fix fuzzy search problems in the web-UI auto-completion.
+
 ## 1.5.2 / 2017-02-10
 
 * [BUGFIX] Fix series corruption in a special case of series maintenance where

MAINTAINERS.md

@@ -1,7 +1,7 @@
 Maintainers of this repository with their focus areas:
 
-* Björn Rabenstein <beorn@soundcloud.com>: Local storage; general code-level issues.
+* Björn Rabenstein <beorn@soundcloud.com> @beorn7: Local storage; general code-level issues.
-* Brian Brazil <brian.brazil@robustperception.io>: Console templates; semantics of PromQL, service discovery, and relabeling.
+* Brian Brazil <brian.brazil@robustperception.io> @brian-brazil: Console templates; semantics of PromQL, service discovery, and relabeling.
-* Fabian Reinartz <fabian.reinartz@coreos.com>: PromQL parsing and evaluation; implementation of retrieval, alert notification, and service discovery.
+* Fabian Reinartz <fabian.reinartz@coreos.com> @fabxc: PromQL parsing and evaluation; implementation of retrieval, alert notification, and service discovery.
-* Julius Volz <julius.volz@gmail.com>: Remote storage integrations; web UI.
+* Julius Volz <julius.volz@gmail.com> @juliusv: Remote storage integrations; web UI.
 

Makefile

@@ -37,9 +37,13 @@ check_license:
 	@./scripts/check_license.sh
 
 # TODO(fabxc): example tests temporarily removed.
-test:
+test-short:
 	@echo ">> running short tests"
-	@$(GO) test -short $(shell $(GO) list ./... | grep -v /vendor/ | grep -v examples)
+	@$(GO) test $(shell $(GO) list ./... | grep -v /vendor/ | grep -v examples)
+
+test:
+	@echo ">> running all tests"
+	@$(GO) test $(pkgs)
 
 format:
 	@echo ">> formatting code"

README.md

@@ -81,6 +81,7 @@ The Makefile provides several targets:
 
   * *build*: build the `prometheus` and `promtool` binaries
   * *test*: run the tests
+  * *test-short*: run the short tests
   * *format*: format the source code
   * *vet*: check the source code for common errors
   * *assets*: rebuild the static assets

circle.yml

@@ -49,6 +49,7 @@ deployment:
     owner: prometheus
     commands:
       - promu crossbuild tarballs
+      - promu checksum .tarballs
       - promu release .tarballs
       - mkdir $CIRCLE_ARTIFACTS/releases/ && cp -a .tarballs/* $CIRCLE_ARTIFACTS/releases/
       - docker login -e $DOCKER_EMAIL -u $DOCKER_LOGIN -p $DOCKER_PASSWORD

cmd/promtool/main.go

@@ -24,6 +24,7 @@ import (
 	"github.com/prometheus/prometheus/config"
 	"github.com/prometheus/prometheus/promql"
 	"github.com/prometheus/prometheus/util/cli"
+	"github.com/prometheus/prometheus/util/promlint"
 )
 
 // CheckConfigCmd validates configuration files.
@@ -182,6 +183,42 @@ func checkRules(t cli.Term, filename string) (int, error) {
 	return len(rules), nil
 }
 
+var checkMetricsUsage = strings.TrimSpace(`
+usage: promtool check-metrics
+
+Pass Prometheus metrics over stdin to lint them for consistency and correctness.
+
+examples:
+
+$ cat metrics.prom | promtool check-metrics
+$ curl -s http://localhost:9090/metrics | promtool check-metrics
+`)
+
+// CheckMetricsCmd performs a linting pass on input metrics.
+func CheckMetricsCmd(t cli.Term, args ...string) int {
+	if len(args) != 0 {
+		t.Infof(checkMetricsUsage)
+		return 2
+	}
+
+	l := promlint.New(os.Stdin)
+	problems, err := l.Lint()
+	if err != nil {
+		t.Errorf("error while linting: %v", err)
+		return 1
+	}
+
+	for _, p := range problems {
+		t.Errorf("%s: %s", p.Metric, p.Text)
+	}
+
+	if len(problems) > 0 {
+		return 3
+	}
+
+	return 0
+}
+
 // VersionCmd prints the binaries version information.
 func VersionCmd(t cli.Term, _ ...string) int {
 	fmt.Fprintln(os.Stdout, version.Print("promtool"))
@@ -201,6 +238,11 @@ func main() {
 		Run:  CheckRulesCmd,
 	})
 
+	app.Register("check-metrics", &cli.Command{
+		Desc: "validate metrics for correctness",
+		Run:  CheckMetricsCmd,
+	})
+
 	app.Register("version", &cli.Command{
 		Desc: "print the version of this binary",
 		Run:  VersionCmd,

console_libraries/prom.lib

@@ -1,20 +1,19 @@
 {{/* vim: set ft=html: */}}
 {{/* Load Prometheus console library JS/CSS. Should go in <head> */}}
 {{ define "prom_console_head" }}
-<link type="text/css" rel="stylesheet" href="{{ pathPrefix }}/static/vendor/rickshaw/rickshaw.min.css?v={{ buildVersion }}">
+<link type="text/css" rel="stylesheet" href="{{ pathPrefix }}/static/vendor/rickshaw/rickshaw.min.css">
-<link type="text/css" rel="stylesheet" href="{{ pathPrefix }}/static/vendor/bootstrap-3.3.1/css/bootstrap.min.css?v={{ buildVersion }}">
+<link type="text/css" rel="stylesheet" href="{{ pathPrefix }}/static/vendor/bootstrap-3.3.1/css/bootstrap.min.css">
-<link type="text/css" rel="stylesheet" href="{{ pathPrefix }}/static/css/prom_console.css?v={{ buildVersion }}">
+<link type="text/css" rel="stylesheet" href="{{ pathPrefix }}/static/css/prom_console.css">
-<script src="{{ pathPrefix }}/static/vendor/rickshaw/vendor/d3.v3.js?v={{ buildVersion }}"></script>
+<script src="{{ pathPrefix }}/static/vendor/rickshaw/vendor/d3.v3.js"></script>
-<script src="{{ pathPrefix }}/static/vendor/rickshaw/vendor/d3.layout.min.js?v={{ buildVersion }}"></script>
+<script src="{{ pathPrefix }}/static/vendor/rickshaw/vendor/d3.layout.min.js"></script>
-<script src="{{ pathPrefix }}/static/vendor/rickshaw/rickshaw.min.js?v={{ buildVersion }}"></script>
+<script src="{{ pathPrefix }}/static/vendor/rickshaw/rickshaw.min.js"></script>
-<script src="{{ pathPrefix }}/static/vendor/js/jquery.min.js?v={{ buildVersion }}"></script>
+<script src="{{ pathPrefix }}/static/vendor/js/jquery.min.js"></script>
-<script src="{{ pathPrefix }}/static/vendor/bootstrap-3.3.1/js/bootstrap.min.js?v={{ buildVersion }}"></script>
+<script src="{{ pathPrefix }}/static/vendor/bootstrap-3.3.1/js/bootstrap.min.js"></script>
 
 <script>
 var PATH_PREFIX = "{{ pathPrefix }}";
-var BUILD_VERSION = "{{ buildVersion }}";
 </script>
-<script src="{{ pathPrefix }}/static/js/prom_console.js?v={{ buildVersion }}"></script>
+<script src="{{ pathPrefix }}/static/js/prom_console.js"></script>
 {{ end }}
 
 {{/* Top of all pages. */}}

discovery/discovery.go

@@ -217,11 +217,6 @@ func (ts *TargetSet) UpdateProviders(p map[string]TargetProvider) {
 }
 
 func (ts *TargetSet) updateProviders(ctx context.Context, providers map[string]TargetProvider) {
-	// Lock for the entire time. This may mean up to 5 seconds until the full initial set
-	// is retrieved and applied.
-	// We could release earlier with some tweaks, but this is easier to reason about.
-	ts.mtx.Lock()
-	defer ts.mtx.Unlock()
 
 	// Stop all previous target providers of the target set.
 	if ts.cancelProviders != nil {
@@ -233,7 +228,9 @@ func (ts *TargetSet) updateProviders(ctx context.Context, providers map[string]T
 	// (Re-)create a fresh tgroups map to not keep stale targets around. We
 	// will retrieve all targets below anyway, so cleaning up everything is
 	// safe and doesn't inflict any additional cost.
+	ts.mtx.Lock()
 	ts.tgroups = map[string]*config.TargetGroup{}
+	ts.mtx.Unlock()
 
 	for name, prov := range providers {
 		wg.Add(1)
@@ -292,9 +289,6 @@ func (ts *TargetSet) updateProviders(ctx context.Context, providers map[string]T
 
 // update handles a target group update from a target provider identified by the name.
 func (ts *TargetSet) update(name string, tgroup *config.TargetGroup) {
-	ts.mtx.Lock()
-	defer ts.mtx.Unlock()
-
 	ts.setTargetGroup(name, tgroup)
 
 	select {
@@ -304,6 +298,9 @@ func (ts *TargetSet) update(name string, tgroup *config.TargetGroup) {
 }
 
 func (ts *TargetSet) setTargetGroup(name string, tg *config.TargetGroup) {
+	ts.mtx.Lock()
+	defer ts.mtx.Unlock()
+
 	if tg == nil {
 		return
 	}

discovery/triton/triton.go

@@ -32,6 +32,7 @@ const (
 	tritonLabel             = model.MetaLabelPrefix + "triton_"
 	tritonLabelMachineId    = tritonLabel + "machine_id"
 	tritonLabelMachineAlias = tritonLabel + "machine_alias"
+	tritonLabelMachineBrand = tritonLabel + "machine_brand"
 	tritonLabelMachineImage = tritonLabel + "machine_image"
 	tritonLabelServerId     = tritonLabel + "server_id"
 	namespace               = "prometheus"
@@ -59,6 +60,7 @@ type DiscoveryResponse struct {
 	Containers []struct {
 		ServerUUID  string `json:"server_uuid"`
 		VMAlias     string `json:"vm_alias"`
+		VMBrand     string `json:"vm_brand"`
 		VMImageUUID string `json:"vm_image_uuid"`
 		VMUUID      string `json:"vm_uuid"`
 	} `json:"containers"`
@@ -157,6 +159,7 @@ func (d *Discovery) refresh() (tg *config.TargetGroup, err error) {
 		labels := model.LabelSet{
 			tritonLabelMachineId:    model.LabelValue(container.VMUUID),
 			tritonLabelMachineAlias: model.LabelValue(container.VMAlias),
+			tritonLabelMachineBrand: model.LabelValue(container.VMBrand),
 			tritonLabelMachineImage: model.LabelValue(container.VMImageUUID),
 			tritonLabelServerId:     model.LabelValue(container.ServerUUID),
 		}

discovery/triton/triton_test.go

@@ -111,12 +111,14 @@ func TestTritonSDRefreshMultipleTargets(t *testing.T) {
 		 	{
 				"server_uuid":"44454c4c-5000-104d-8037-b7c04f5a5131",
 				"vm_alias":"server01",
+				"vm_brand":"lx",
 				"vm_image_uuid":"7b27a514-89d7-11e6-bee6-3f96f367bee7",
 				"vm_uuid":"ad466fbf-46a2-4027-9b64-8d3cdb7e9072"
 			},
 			{
 				"server_uuid":"a5894692-bd32-4ca1-908a-e2dda3c3a5e6",
 				"vm_alias":"server02",
+				"vm_brand":"kvm",
 				"vm_image_uuid":"a5894692-bd32-4ca1-908a-e2dda3c3a5e6",
 				"vm_uuid":"7b27a514-89d7-11e6-bee6-3f96f367bee7"
 			}]

documentation/examples/remote_storage/example_write_adapter/README.md

@@ -1,4 +1,4 @@
-## Generic Remote Storage Example
+## Remote Write Adapter Example
 
 This is a simple example of how to write a server to
 receive samples from the remote storage output.
@@ -7,7 +7,7 @@ To use it:
 
 ```
 go build
-./example_receiver
+./example_write_adapter
 ```
 
 ...and then add the following to your `prometheus.yml`:

documentation/examples/remote_storage/remote_storage_adapter/README.md

@@ -0,0 +1,55 @@
+# Remote storage adapter
+
+This is a write adapter that receives samples via Prometheus's remote write
+protocol and stores them in Graphite, InfluxDB, or OpenTSDB. It is meant as a
+replacement for the built-in specific remote storage implementations that have
+been removed from Prometheus.
+
+For InfluxDB, this binary is also a read adapter that supports reading back
+data through Prometheus via Prometheus's remote read protocol.
+
+## Building
+
+```
+go build
+```
+
+## Running
+
+Graphite example:
+
+```
+./remote_storage_adapter -graphite-address=localhost:8080
+```
+
+OpenTSDB example:
+
+```
+./remote_storage_adapter -opentsdb-url=http://localhost:8081/
+```
+
+InfluxDB example:
+
+```
+./remote_storage_adapter -influxdb-url=http://localhost:8086/ -influxdb.database=prometheus -influxdb.retention-policy=autogen
+```
+
+To show all flags:
+
+```
+./remote_storage_adapter -h
+```
+
+## Configuring Prometheus
+
+To configure Prometheus to send samples to this binary, add the following to your `prometheus.yml`:
+
+```yaml
+# Remote write configuration (for Graphite, OpenTSDB, or InfluxDB).
+remote_write:
+  - url: "http://localhost:9201/write"
+
+# Remote read configuration (for InfluxDB only at the moment).
+remote_read:
+  - url: "http://localhost:9201/read"
+```

documentation/examples/remote_storage/remote_storage_adapter/influxdb/client.go

@@ -124,10 +124,12 @@ func (c *Client) Read(req *remote.ReadRequest) (*remote.ReadResponse, error) {
 	}
 
 	resp := remote.ReadResponse{
-		Timeseries: make([]*remote.TimeSeries, 0, len(labelsToSeries)),
+		Results: []*remote.QueryResult{
+			{Timeseries: make([]*remote.TimeSeries, 0, len(labelsToSeries))},
+		},
 	}
 	for _, ts := range labelsToSeries {
-		resp.Timeseries = append(resp.Timeseries, ts)
+		resp.Results[0].Timeseries = append(resp.Results[0].Timeseries, ts)
 	}
 	return &resp, nil
 }

documentation/examples/remote_storage/remote_storage_adapter/main.go

@@ -33,9 +33,9 @@ import (
 
 	influx "github.com/influxdata/influxdb/client/v2"
 
-	"github.com/prometheus/prometheus/documentation/examples/remote_storage/remote_storage_bridge/graphite"
+	"github.com/prometheus/prometheus/documentation/examples/remote_storage/remote_storage_adapter/graphite"
-	"github.com/prometheus/prometheus/documentation/examples/remote_storage/remote_storage_bridge/influxdb"
+	"github.com/prometheus/prometheus/documentation/examples/remote_storage/remote_storage_adapter/influxdb"
-	"github.com/prometheus/prometheus/documentation/examples/remote_storage/remote_storage_bridge/opentsdb"
+	"github.com/prometheus/prometheus/documentation/examples/remote_storage/remote_storage_adapter/opentsdb"
 	"github.com/prometheus/prometheus/storage/remote"
 )
 

documentation/examples/remote_storage/remote_storage_bridge/README.md

@@ -1,55 +0,0 @@
-# Remote storage bridge
-
-This is a bridge that receives samples via Prometheus's remote write
-protocol and stores them in Graphite, InfluxDB, or OpenTSDB. It is meant
-as a replacement for the built-in specific remote storage implementations
-that have been removed from Prometheus.
-
-For InfluxDB, this bridge also supports reading back data through
-Prometheus via Prometheus's remote read protocol.
-
-## Building
-
-```
-go build
-```
-
-## Running
-
-Graphite example:
-
-```
-./remote_storage_bridge -graphite-address=localhost:8080
-```
-
-OpenTSDB example:
-
-```
-./remote_storage_bridge -opentsdb-url=http://localhost:8081/
-```
-
-InfluxDB example:
-
-```
-./remote_storage_bridge -influxdb-url=http://localhost:8086/ -influxdb.database=prometheus -influxdb.retention-policy=autogen
-```
-
-To show all flags:
-
-```
-./remote_storage_bridge -h
-```
-
-## Configuring Prometheus
-
-To configure Prometheus to send samples to this bridge, add the following to your `prometheus.yml`:
-
-```yaml
-# Remote write configuration (for Graphite, OpenTSDB, or InfluxDB).
-remote_write:
-  - url: "http://localhost:9201/write"
-
-# Remote read configuration (for InfluxDB only at the moment).
-remote_read:
-  - url: "http://localhost:9201/read"
-```

notifier/notifier.go

@@ -369,13 +369,13 @@ func (n *Notifier) setMore() {
 	}
 }
 
-// Alertmanagers returns a list Alertmanager URLs.
+// Alertmanagers returns a slice of Alertmanager URLs.
-func (n *Notifier) Alertmanagers() []string {
+func (n *Notifier) Alertmanagers() []*url.URL {
 	n.mtx.RLock()
 	amSets := n.alertmanagers
 	n.mtx.RUnlock()
 
-	var res []string
+	var res []*url.URL
 
 	for _, ams := range amSets {
 		ams.mtx.RLock()
@@ -417,7 +417,7 @@ func (n *Notifier) sendAll(alerts ...*Alert) bool {
 			defer cancel()
 
 			go func(am alertmanager) {
-				u := am.url()
+				u := am.url().String()
 
 				if err := n.sendOne(ctx, ams.client, u, b); err != nil {
 					log.With("alertmanager", u).With("count", len(alerts)).Errorf("Error sending alerts: %s", err)
@@ -465,20 +465,19 @@ func (n *Notifier) Stop() {
 
 // alertmanager holds Alertmanager endpoint information.
 type alertmanager interface {
-	url() string
+	url() *url.URL
 }
 
 type alertmanagerLabels struct{ labels.Labels }
 
 const pathLabel = "__alerts_path__"
 
-func (a alertmanagerLabels) url() string {
+func (a alertmanagerLabels) url() *url.URL {
-	u := &url.URL{
+	return &url.URL{
 		Scheme: a.Get(model.SchemeLabel),
 		Host:   a.Get(model.AddressLabel),
 		Path:   a.Get(pathLabel),
 	}
-	return u.String()
 }
 
 // alertmanagerSet contains a set of Alertmanagers discovered via a group of service
@@ -529,7 +528,7 @@ func (s *alertmanagerSet) Sync(tgs []*config.TargetGroup) {
 	seen := map[string]struct{}{}
 
 	for _, am := range all {
-		us := am.url()
+		us := am.url().String()
 		if _, ok := seen[us]; ok {
 			continue
 		}

notifier/notifier_test.go

@@ -19,6 +19,7 @@ import (
 	"io/ioutil"
 	"net/http"
 	"net/http/httptest"
+	"net/url"
 	"testing"
 	"time"
 
@@ -388,6 +389,10 @@ type alertmanagerMock struct {
 	urlf func() string
 }
 
-func (a alertmanagerMock) url() string {
+func (a alertmanagerMock) url() *url.URL {
-	return a.urlf()
+	u, err := url.Parse(a.urlf())
+	if err != nil {
+		panic(err)
+	}
+	return u
 }

promql/engine.go

@@ -113,6 +113,9 @@ type (
 	ErrQueryTimeout string
 	// ErrQueryCanceled is returned if a query was canceled during processing.
 	ErrQueryCanceled string
+	// ErrStorage is returned if an error was encountered in the storage layer
+	// during query handling.
+	ErrStorage error
 )
 
 func (e ErrQueryTimeout) Error() string  { return fmt.Sprintf("query timed out in %s", string(e)) }

promql/quantile.go

@@ -83,6 +83,8 @@ func bucketQuantile(q float64, buckets buckets) float64 {
 		return math.NaN()
 	}
 
+	ensureMonotonic(buckets)
+
 	rank := q * buckets[len(buckets)-1].count
 	b := sort.Search(len(buckets)-1, func(i int) bool { return buckets[i].count >= rank })
 
@@ -105,7 +107,52 @@ func bucketQuantile(q float64, buckets buckets) float64 {
 	return bucketStart + (bucketEnd-bucketStart)*float64(rank/count)
 }
 
-// qauntile calculates the given quantile of a Vector of samples.
+// The assumption that bucket counts increase monotonically with increasing
+// upperBound may be violated during:
+//
+//   * Recording rule evaluation of histogram_quantile, especially when rate()
+//      has been applied to the underlying bucket timeseries.
+//   * Evaluation of histogram_quantile computed over federated bucket
+//      timeseries, especially when rate() has been applied.
+//
+// This is because scraped data is not made available to rule evaluation or
+// federation atomically, so some buckets are computed with data from the
+// most recent scrapes, but the other buckets are missing data from the most
+// recent scrape.
+//
+// Monotonicity is usually guaranteed because if a bucket with upper bound
+// u1 has count c1, then any bucket with a higher upper bound u > u1 must
+// have counted all c1 observations and perhaps more, so that c  >= c1.
+//
+// Randomly interspersed partial sampling breaks that guarantee, and rate()
+// exacerbates it. Specifically, suppose bucket le=1000 has a count of 10 from
+// 4 samples but the bucket with le=2000 has a count of 7 from 3 samples. The
+// monotonicity is broken. It is exacerbated by rate() because under normal
+// operation, cumulative counting of buckets will cause the bucket counts to
+// diverge such that small differences from missing samples are not a problem.
+// rate() removes this divergence.)
+//
+// bucketQuantile depends on that monotonicity to do a binary search for the
+// bucket with the φ-quantile count, so breaking the monotonicity
+// guarantee causes bucketQuantile() to return undefined (nonsense) results.
+//
+// As a somewhat hacky solution until ingestion is atomic per scrape, we
+// calculate the "envelope" of the histogram buckets, essentially removing
+// any decreases in the count between successive buckets.
+
+func ensureMonotonic(buckets buckets) {
+	max := buckets[0].count
+	for i := range buckets[1:] {
+		switch {
+		case buckets[i].count > max:
+			max = buckets[i].count
+		case buckets[i].count < max:
+			buckets[i].count = max
+		}
+	}
+}
+
+// qauntile calculates the given quantile of a vector of samples.
 //
 // The Vector will be sorted.
 // If 'values' has zero elements, NaN is returned.

promql/testdata/histograms.test

@@ -139,3 +139,21 @@ eval instant at 50m histogram_quantile(0.5, rate(request_duration_seconds_bucket
 	{instance="ins2", job="job1"} 0.13333333333333333
 	{instance="ins1", job="job2"} 0.1
 	{instance="ins2", job="job2"} 0.11666666666666667
+
+# A histogram with nonmonotonic bucket counts. This may happen when recording
+# rule evaluation or federation races scrape ingestion, causing some buckets
+# counts to be derived from fewer samples. The wrong answer we want to avoid
+# is for histogram_quantile(0.99, nonmonotonic_bucket) to return ~1000 instead
+# of 1.
+
+load 5m
+    nonmonotonic_bucket{le="0.1"}   0+1x10
+    nonmonotonic_bucket{le="1"}     0+9x10
+    nonmonotonic_bucket{le="10"}    0+8x10
+    nonmonotonic_bucket{le="100"}   0+8x10
+    nonmonotonic_bucket{le="1000"}  0+9x10
+    nonmonotonic_bucket{le="+Inf"}  0+9x10
+
+# Nonmonotonic buckets
+eval instant at 50m histogram_quantile(0.99, nonmonotonic_bucket)
+    {} 0.989875

retrieval/scrape.go

@@ -184,7 +184,7 @@ func (sp *scrapePool) reload(cfg *config.ScrapeConfig) {
 	for fp, oldLoop := range sp.loops {
 		var (
 			t       = sp.targets[fp]
-			s       = &targetScraper{Target: t, client: sp.client}
+			s       = &targetScraper{Target: t, client: sp.client, timeout: timeout}
 			newLoop = sp.newLoop(sp.ctx, s,
 				func() storage.Appender {
 					return sp.sampleAppender(t)
@@ -253,7 +253,7 @@ func (sp *scrapePool) sync(targets []*Target) {
 		uniqueTargets[hash] = struct{}{}
 
 		if _, ok := sp.targets[hash]; !ok {
-			s := &targetScraper{Target: t, client: sp.client}
+			s := &targetScraper{Target: t, client: sp.client, timeout: timeout}
 			l := sp.newLoop(sp.ctx, s,
 				func() storage.Appender {
 					return sp.sampleAppender(t)
@@ -354,8 +354,9 @@ type scraper interface {
 type targetScraper struct {
 	*Target
 
-	client *http.Client
+	client  *http.Client
-	req    *http.Request
+	req     *http.Request
+	timeout time.Duration
 
 	gzipr *gzip.Reader
 	buf   *bufio.Reader
@@ -372,13 +373,13 @@ func (s *targetScraper) scrape(ctx context.Context, w io.Writer) error {
 			return err
 		}
 		// Disable accept header to always negotiate for text format.
-		// req.Header.Add("Accept", acceptHeader)
+		req.Header.Add("Accept", acceptHeader)
 		req.Header.Add("Accept-Encoding", "gzip")
 		req.Header.Set("User-Agent", userAgentHeader)
+		req.Header.Set("X-Prometheus-Scrape-Timeout-Seconds", fmt.Sprintf("%f", s.timeout.Seconds()))
 
 		s.req = req
 	}
-
 	resp, err := ctxhttp.Do(ctx, s.client, s.req)
 	if err != nil {
 		return err

retrieval/scrape_test.go

@@ -435,8 +435,20 @@ func TestScrapeLoopRun(t *testing.T) {
 }
 
 func TestTargetScraperScrapeOK(t *testing.T) {
+	const (
+		configTimeout   = 1500 * time.Millisecond
+		expectedTimeout = "1.500000"
+	)
+
 	server := httptest.NewServer(
 		http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+			timeout := r.Header.Get("X-Prometheus-Scrape-Timeout-Seconds")
+			if timeout != expectedTimeout {
+				t.Errorf("Scrape timeout did not match expected timeout")
+				t.Errorf("Expected: %v", expectedTimeout)
+				t.Fatalf("Got: %v", timeout)
+			}
+
 			w.Header().Set("Content-Type", `text/plain; version=0.0.4`)
 			w.Write([]byte("metric_a 1\nmetric_b 2\n"))
 		}),
@@ -455,7 +467,8 @@ func TestTargetScraperScrapeOK(t *testing.T) {
 				model.AddressLabel, serverURL.Host,
 			),
 		},
-		client: http.DefaultClient,
+		client:  http.DefaultClient,
+		timeout: configTimeout,
 	}
 	var buf bytes.Buffer
 

util/promlint/promlint.go

@@ -0,0 +1,268 @@
+// Copyright 2017 The Prometheus Authors
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+// Package promlint provides a linter for Prometheus metrics.
+package promlint
+
+import (
+	"fmt"
+	"io"
+	"sort"
+	"strings"
+
+	dto "github.com/prometheus/client_model/go"
+	"github.com/prometheus/common/expfmt"
+)
+
+// A Linter is a Prometheus metrics linter.  It identifies issues with metric
+// names, types, and metadata, and reports them to the caller.
+type Linter struct {
+	r io.Reader
+}
+
+// A Problem is an issue detected by a Linter.
+type Problem struct {
+	// The name of the metric indicated by this Problem.
+	Metric string
+
+	// A description of the issue for this Problem.
+	Text string
+}
+
+// problems is a slice of Problems with a helper method to easily append
+// additional Problems to the slice.
+type problems []Problem
+
+// Add appends a new Problem to the slice for the specified metric, with
+// the specified issue text.
+func (p *problems) Add(mf dto.MetricFamily, text string) {
+	*p = append(*p, Problem{
+		Metric: mf.GetName(),
+		Text:   text,
+	})
+}
+
+// New creates a new Linter that reads an input stream of Prometheus metrics.
+// Only the text exposition format is supported.
+func New(r io.Reader) *Linter {
+	return &Linter{
+		r: r,
+	}
+}
+
+// Lint performs a linting pass, returning a slice of Problems indicating any
+// issues found in the metrics stream.  The slice is sorted by metric name
+// and issue description.
+func (l *Linter) Lint() ([]Problem, error) {
+	// TODO(mdlayher): support for protobuf exposition format?
+	d := expfmt.NewDecoder(l.r, expfmt.FmtText)
+
+	var problems []Problem
+
+	var mf dto.MetricFamily
+	for {
+		if err := d.Decode(&mf); err != nil {
+			if err == io.EOF {
+				break
+			}
+
+			return nil, err
+		}
+
+		problems = append(problems, lint(mf)...)
+	}
+
+	// Ensure deterministic output.
+	sort.SliceStable(problems, func(i, j int) bool {
+		if problems[i].Metric < problems[j].Metric {
+			return true
+		}
+
+		return problems[i].Text < problems[j].Text
+	})
+
+	return problems, nil
+}
+
+// lint is the entry point for linting a single metric.
+func lint(mf dto.MetricFamily) []Problem {
+	fns := []func(mf dto.MetricFamily) []Problem{
+		lintHelp,
+		lintMetricUnits,
+		lintCounter,
+		lintHistogramSummaryReserved,
+	}
+
+	var problems []Problem
+	for _, fn := range fns {
+		problems = append(problems, fn(mf)...)
+	}
+
+	// TODO(mdlayher): lint rules for specific metrics types.
+	return problems
+}
+
+// lintHelp detects issues related to the help text for a metric.
+func lintHelp(mf dto.MetricFamily) []Problem {
+	var problems problems
+
+	// Expect all metrics to have help text available.
+	if mf.Help == nil {
+		problems.Add(mf, "no help text")
+	}
+
+	return problems
+}
+
+// lintMetricUnits detects issues with metric unit names.
+func lintMetricUnits(mf dto.MetricFamily) []Problem {
+	var problems problems
+
+	unit, base, ok := metricUnits(*mf.Name)
+	if !ok {
+		// No known units detected.
+		return nil
+	}
+
+	// Unit is already a base unit.
+	if unit == base {
+		return nil
+	}
+
+	problems.Add(mf, fmt.Sprintf("use base unit %q instead of %q", base, unit))
+
+	return problems
+}
+
+// lintCounter detects issues specific to counters, as well as patterns that should
+// only be used with counters.
+func lintCounter(mf dto.MetricFamily) []Problem {
+	var problems problems
+
+	isCounter := mf.GetType() == dto.MetricType_COUNTER
+	isUntyped := mf.GetType() == dto.MetricType_UNTYPED
+	hasTotalSuffix := strings.HasSuffix(mf.GetName(), "_total")
+
+	switch {
+	case isCounter && !hasTotalSuffix:
+		problems.Add(mf, `counter metrics should have "_total" suffix`)
+	case !isUntyped && !isCounter && hasTotalSuffix:
+		problems.Add(mf, `non-counter metrics should not have "_total" suffix`)
+	}
+
+	return problems
+}
+
+// lintHistogramSummaryReserved detects when other types of metrics use names or labels
+// reserved for use by histograms and/or summaries.
+func lintHistogramSummaryReserved(mf dto.MetricFamily) []Problem {
+	// These rules do not apply to untyped metrics.
+	t := mf.GetType()
+	if t == dto.MetricType_UNTYPED {
+		return nil
+	}
+
+	var problems problems
+
+	isHistogram := t == dto.MetricType_HISTOGRAM
+	isSummary := t == dto.MetricType_SUMMARY
+
+	n := mf.GetName()
+
+	if !isHistogram && strings.HasSuffix(n, "_bucket") {
+		problems.Add(mf, `non-histogram metrics should not have "_bucket" suffix`)
+	}
+	if !isHistogram && !isSummary && strings.HasSuffix(n, "_count") {
+		problems.Add(mf, `non-histogram and non-summary metrics should not have "_count" suffix`)
+	}
+	if !isHistogram && !isSummary && strings.HasSuffix(n, "_sum") {
+		problems.Add(mf, `non-histogram and non-summary metrics should not have "_sum" suffix`)
+	}
+
+	for _, m := range mf.GetMetric() {
+		for _, l := range m.GetLabel() {
+			ln := l.GetName()
+
+			if !isHistogram && ln == "le" {
+				problems.Add(mf, `non-histogram metrics should not have "le" label`)
+			}
+			if !isSummary && ln == "quantile" {
+				problems.Add(mf, `non-summary metrics should not have "quantile" label`)
+			}
+		}
+	}
+
+	return problems
+}
+
+// metricUnits attempts to detect known unit types used as part of a metric name,
+// e.g. "foo_bytes_total" or "bar_baz_milligrams".
+func metricUnits(m string) (unit string, base string, ok bool) {
+	ss := strings.Split(m, "_")
+
+	for _, u := range baseUnits {
+		// Also check for "no prefix".
+		for _, p := range append(unitPrefixes, "") {
+			for _, s := range ss {
+				// Attempt to explicitly match a known unit with a known prefix,
+				// as some words may look like "units" when matching suffix.
+				//
+				// As an example, "thermometers" should not match "meters", but
+				// "kilometers" should.
+				if s == p+u {
+					return p + u, u, true
+				}
+			}
+		}
+	}
+
+	return "", "", false
+}
+
+// Units and their possible prefixes recognized by this library.  More can be
+// added over time as needed.
+var (
+	baseUnits = []string{
+		"amperes",
+		"bytes",
+		"candela",
+		"grams",
+		"kelvin", // Both plural and non-plural form allowed.
+		"kelvins",
+		"meters", // Both American and international spelling permitted.
+		"metres",
+		"moles",
+		"seconds",
+	}
+
+	unitPrefixes = []string{
+		"pico",
+		"nano",
+		"micro",
+		"milli",
+		"centi",
+		"deci",
+		"deca",
+		"hecto",
+		"kilo",
+		"kibi",
+		"mega",
+		"mibi",
+		"giga",
+		"gibi",
+		"tera",
+		"tebi",
+		"peta",
+		"pebi",
+	}
+)

util/promlint/promlint_test.go

@@ -0,0 +1,497 @@
+// Copyright 2017 The Prometheus Authors
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package promlint_test
+
+import (
+	"reflect"
+	"strings"
+	"testing"
+
+	"github.com/prometheus/prometheus/util/promlint"
+)
+
+func TestLintNoHelpText(t *testing.T) {
+	const msg = "no help text"
+
+	tests := []struct {
+		name     string
+		in       string
+		problems []promlint.Problem
+	}{
+		{
+			name: "no help",
+			in: `
+# TYPE go_goroutines gauge
+go_goroutines 24
+`,
+			problems: []promlint.Problem{{
+				Metric: "go_goroutines",
+				Text:   msg,
+			}},
+		},
+		{
+			name: "empty help",
+			in: `
+# HELP go_goroutines
+# TYPE go_goroutines gauge
+go_goroutines 24
+`,
+			problems: []promlint.Problem{{
+				Metric: "go_goroutines",
+				Text:   msg,
+			}},
+		},
+		{
+			name: "no help and empty help",
+			in: `
+# HELP go_goroutines
+# TYPE go_goroutines gauge
+go_goroutines 24
+# TYPE go_threads gauge
+go_threads 10
+`,
+			problems: []promlint.Problem{
+				{
+					Metric: "go_goroutines",
+					Text:   msg,
+				},
+				{
+					Metric: "go_threads",
+					Text:   msg,
+				},
+			},
+		},
+		{
+			name: "OK",
+			in: `
+# HELP go_goroutines Number of goroutines that currently exist.
+# TYPE go_goroutines gauge
+go_goroutines 24
+`,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			l := promlint.New(strings.NewReader(tt.in))
+
+			problems, err := l.Lint()
+			if err != nil {
+				t.Fatalf("unexpected error: %v", err)
+			}
+
+			if want, got := tt.problems, problems; !reflect.DeepEqual(want, got) {
+				t.Fatalf("unexpected problems:\n- want: %v\n-  got: %v",
+					want, got)
+			}
+		})
+	}
+}
+
+func TestLintMetricUnits(t *testing.T) {
+	tests := []struct {
+		name     string
+		in       string
+		problems []promlint.Problem
+	}{
+		{
+			name: "amperes",
+			in: `
+# HELP x_milliamperes Test metric.
+# TYPE x_milliamperes untyped
+x_milliamperes 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_milliamperes",
+				Text:   `use base unit "amperes" instead of "milliamperes"`,
+			}},
+		},
+		{
+			name: "bytes",
+			in: `
+# HELP x_gigabytes Test metric.
+# TYPE x_gigabytes untyped
+x_gigabytes 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_gigabytes",
+				Text:   `use base unit "bytes" instead of "gigabytes"`,
+			}},
+		},
+		{
+			name: "candela",
+			in: `
+# HELP x_kilocandela Test metric.
+# TYPE x_kilocandela untyped
+x_kilocandela 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_kilocandela",
+				Text:   `use base unit "candela" instead of "kilocandela"`,
+			}},
+		},
+		{
+			name: "grams",
+			in: `
+# HELP x_kilograms Test metric.
+# TYPE x_kilograms untyped
+x_kilograms 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_kilograms",
+				Text:   `use base unit "grams" instead of "kilograms"`,
+			}},
+		},
+		{
+			name: "kelvin",
+			in: `
+# HELP x_nanokelvin Test metric.
+# TYPE x_nanokelvin untyped
+x_nanokelvin 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_nanokelvin",
+				Text:   `use base unit "kelvin" instead of "nanokelvin"`,
+			}},
+		},
+		{
+			name: "kelvins",
+			in: `
+# HELP x_nanokelvins Test metric.
+# TYPE x_nanokelvins untyped
+x_nanokelvins 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_nanokelvins",
+				Text:   `use base unit "kelvins" instead of "nanokelvins"`,
+			}},
+		},
+		{
+			name: "meters",
+			in: `
+# HELP x_kilometers Test metric.
+# TYPE x_kilometers untyped
+x_kilometers 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_kilometers",
+				Text:   `use base unit "meters" instead of "kilometers"`,
+			}},
+		},
+		{
+			name: "metres",
+			in: `
+# HELP x_kilometres Test metric.
+# TYPE x_kilometres untyped
+x_kilometres 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_kilometres",
+				Text:   `use base unit "metres" instead of "kilometres"`,
+			}},
+		},
+		{
+			name: "moles",
+			in: `
+# HELP x_picomoles Test metric.
+# TYPE x_picomoles untyped
+x_picomoles 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_picomoles",
+				Text:   `use base unit "moles" instead of "picomoles"`,
+			}},
+		},
+		{
+			name: "seconds",
+			in: `
+# HELP x_microseconds Test metric.
+# TYPE x_microseconds untyped
+x_microseconds 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_microseconds",
+				Text:   `use base unit "seconds" instead of "microseconds"`,
+			}},
+		},
+		{
+			name: "OK",
+			in: `
+# HELP thermometers_kelvin Test metric with name that looks like "meters".
+# TYPE thermometers_kelvin untyped
+thermometers_kelvin 0
+`,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			l := promlint.New(strings.NewReader(tt.in))
+
+			problems, err := l.Lint()
+			if err != nil {
+				t.Fatalf("unexpected error: %v", err)
+			}
+
+			if want, got := tt.problems, problems; !reflect.DeepEqual(want, got) {
+				t.Fatalf("unexpected problems:\n- want: %v\n-  got: %v",
+					want, got)
+			}
+		})
+	}
+}
+
+func TestLintCounter(t *testing.T) {
+	tests := []struct {
+		name     string
+		in       string
+		problems []promlint.Problem
+	}{
+		{
+			name: "counter without _total suffix",
+			in: `
+# HELP x_bytes Test metric.
+# TYPE x_bytes counter
+x_bytes 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_bytes",
+				Text:   `counter metrics should have "_total" suffix`,
+			}},
+		},
+		{
+			name: "gauge with _total suffix",
+			in: `
+# HELP x_bytes_total Test metric.
+# TYPE x_bytes_total gauge
+x_bytes_total 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_bytes_total",
+				Text:   `non-counter metrics should not have "_total" suffix`,
+			}},
+		},
+		{
+			name: "counter with _total suffix",
+			in: `
+# HELP x_bytes_total Test metric.
+# TYPE x_bytes_total counter
+x_bytes_total 10
+`,
+		},
+		{
+			name: "gauge without _total suffix",
+			in: `
+# HELP x_bytes Test metric.
+# TYPE x_bytes gauge
+x_bytes 10
+`,
+		},
+		{
+			name: "untyped with _total suffix",
+			in: `
+# HELP x_bytes_total Test metric.
+# TYPE x_bytes_total untyped
+x_bytes_total 10
+`,
+		},
+		{
+			name: "untyped without _total suffix",
+			in: `
+# HELP x_bytes Test metric.
+# TYPE x_bytes untyped
+x_bytes 10
+`,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			l := promlint.New(strings.NewReader(tt.in))
+
+			problems, err := l.Lint()
+			if err != nil {
+				t.Fatalf("unexpected error: %v", err)
+			}
+
+			if want, got := tt.problems, problems; !reflect.DeepEqual(want, got) {
+				t.Fatalf("unexpected problems:\n- want: %v\n-  got: %v",
+					want, got)
+			}
+		})
+	}
+}
+
+func TestLintHistogramSummaryReserved(t *testing.T) {
+	tests := []struct {
+		name     string
+		in       string
+		problems []promlint.Problem
+	}{
+		{
+			name: "gauge with _bucket suffix",
+			in: `
+# HELP x_bytes_bucket Test metric.
+# TYPE x_bytes_bucket gauge
+x_bytes_bucket 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_bytes_bucket",
+				Text:   `non-histogram metrics should not have "_bucket" suffix`,
+			}},
+		},
+		{
+			name: "gauge with _count suffix",
+			in: `
+# HELP x_bytes_count Test metric.
+# TYPE x_bytes_count gauge
+x_bytes_count 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_bytes_count",
+				Text:   `non-histogram and non-summary metrics should not have "_count" suffix`,
+			}},
+		},
+		{
+			name: "gauge with _sum suffix",
+			in: `
+# HELP x_bytes_sum Test metric.
+# TYPE x_bytes_sum gauge
+x_bytes_sum 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_bytes_sum",
+				Text:   `non-histogram and non-summary metrics should not have "_sum" suffix`,
+			}},
+		},
+		{
+			name: "gauge with le label",
+			in: `
+# HELP x_bytes Test metric.
+# TYPE x_bytes gauge
+x_bytes{le="1"} 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_bytes",
+				Text:   `non-histogram metrics should not have "le" label`,
+			}},
+		},
+		{
+			name: "gauge with quantile label",
+			in: `
+# HELP x_bytes Test metric.
+# TYPE x_bytes gauge
+x_bytes{quantile="1"} 10
+`,
+			problems: []promlint.Problem{{
+				Metric: "x_bytes",
+				Text:   `non-summary metrics should not have "quantile" label`,
+			}},
+		},
+		{
+			name: "histogram with quantile label",
+			in: `
+# HELP tsdb_compaction_duration Duration of compaction runs.
+# TYPE tsdb_compaction_duration histogram
+tsdb_compaction_duration_bucket{le="0.005",quantile="0.01"} 0
+tsdb_compaction_duration_bucket{le="0.01",quantile="0.01"} 0
+tsdb_compaction_duration_bucket{le="0.025",quantile="0.01"} 0
+tsdb_compaction_duration_bucket{le="0.05",quantile="0.01"} 0
+tsdb_compaction_duration_bucket{le="0.1",quantile="0.01"} 0
+tsdb_compaction_duration_bucket{le="0.25",quantile="0.01"} 0
+tsdb_compaction_duration_bucket{le="0.5",quantile="0.01"} 57
+tsdb_compaction_duration_bucket{le="1",quantile="0.01"} 68
+tsdb_compaction_duration_bucket{le="2.5",quantile="0.01"} 69
+tsdb_compaction_duration_bucket{le="5",quantile="0.01"} 69
+tsdb_compaction_duration_bucket{le="10",quantile="0.01"} 69
+tsdb_compaction_duration_bucket{le="+Inf",quantile="0.01"} 69
+tsdb_compaction_duration_sum 28.740810936000006
+tsdb_compaction_duration_count 69
+`,
+			problems: []promlint.Problem{{
+				Metric: "tsdb_compaction_duration",
+				Text:   `non-summary metrics should not have "quantile" label`,
+			}},
+		},
+		{
+			name: "summary with le label",
+			in: `
+# HELP go_gc_duration_seconds A summary of the GC invocation durations.
+# TYPE go_gc_duration_seconds summary
+go_gc_duration_seconds{quantile="0",le="0.01"} 4.2365e-05
+go_gc_duration_seconds{quantile="0.25",le="0.01"} 8.1492e-05
+go_gc_duration_seconds{quantile="0.5",le="0.01"} 0.000100656
+go_gc_duration_seconds{quantile="0.75",le="0.01"} 0.000113913
+go_gc_duration_seconds{quantile="1",le="0.01"} 0.021754305
+go_gc_duration_seconds_sum 1.769429004
+go_gc_duration_seconds_count 5962
+`,
+			problems: []promlint.Problem{{
+				Metric: "go_gc_duration_seconds",
+				Text:   `non-histogram metrics should not have "le" label`,
+			}},
+		},
+		{
+			name: "histogram OK",
+			in: `
+# HELP tsdb_compaction_duration Duration of compaction runs.
+# TYPE tsdb_compaction_duration histogram
+tsdb_compaction_duration_bucket{le="0.005"} 0
+tsdb_compaction_duration_bucket{le="0.01"} 0
+tsdb_compaction_duration_bucket{le="0.025"} 0
+tsdb_compaction_duration_bucket{le="0.05"} 0
+tsdb_compaction_duration_bucket{le="0.1"} 0
+tsdb_compaction_duration_bucket{le="0.25"} 0
+tsdb_compaction_duration_bucket{le="0.5"} 57
+tsdb_compaction_duration_bucket{le="1"} 68
+tsdb_compaction_duration_bucket{le="2.5"} 69
+tsdb_compaction_duration_bucket{le="5"} 69
+tsdb_compaction_duration_bucket{le="10"} 69
+tsdb_compaction_duration_bucket{le="+Inf"} 69
+tsdb_compaction_duration_sum 28.740810936000006
+tsdb_compaction_duration_count 69
+`,
+		},
+		{
+			name: "summary OK",
+			in: `
+# HELP go_gc_duration_seconds A summary of the GC invocation durations.
+# TYPE go_gc_duration_seconds summary
+go_gc_duration_seconds{quantile="0"} 4.2365e-05
+go_gc_duration_seconds{quantile="0.25"} 8.1492e-05
+go_gc_duration_seconds{quantile="0.5"} 0.000100656
+go_gc_duration_seconds{quantile="0.75"} 0.000113913
+go_gc_duration_seconds{quantile="1"} 0.021754305
+go_gc_duration_seconds_sum 1.769429004
+go_gc_duration_seconds_count 5962
+`,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			l := promlint.New(strings.NewReader(tt.in))
+
+			problems, err := l.Lint()
+			if err != nil {
+				t.Fatalf("unexpected error: %v", err)
+			}
+
+			if want, got := tt.problems, problems; !reflect.DeepEqual(want, got) {
+				t.Fatalf("unexpected problems:\n- want: %v\n-  got: %v",
+					want, got)
+			}
+		})
+	}
+}

vendor/github.com/prometheus/client_golang/prometheus/counter.go

@@ -30,16 +30,8 @@ type Counter interface {
 	Metric
 	Collector
 
-	// Set is used to set the Counter to an arbitrary value. It is only used
+	// Inc increments the counter by 1. Use Add to increment it by arbitrary
-	// if you have to transfer a value from an external counter into this
+	// non-negative values.
-	// Prometheus metric. Do not use it for regular handling of a
-	// Prometheus counter (as it can be used to break the contract of
-	// monotonically increasing values).
-	//
-	// Deprecated: Use NewConstMetric to create a counter for an external
-	// value. A Counter should never be set.
-	Set(float64)
-	// Inc increments the counter by 1.
 	Inc()
 	// Add adds the given value to the counter. It panics if the value is <
 	// 0.

vendor/github.com/prometheus/client_golang/prometheus/desc.go

@@ -16,20 +16,15 @@ package prometheus
 import (
 	"errors"
 	"fmt"
-	"regexp"
 	"sort"
 	"strings"
 
 	"github.com/golang/protobuf/proto"
+	"github.com/prometheus/common/model"
 
 	dto "github.com/prometheus/client_model/go"
 )
 
-var (
-	metricNameRE = regexp.MustCompile(`^[a-zA-Z_][a-zA-Z0-9_:]*$`)
-	labelNameRE  = regexp.MustCompile("^[a-zA-Z_][a-zA-Z0-9_]*$")
-)
-
 // reservedLabelPrefix is a prefix which is not legal in user-supplied
 // label names.
 const reservedLabelPrefix = "__"
@@ -78,7 +73,7 @@ type Desc struct {
 	// Help string. Each Desc with the same fqName must have the same
 	// dimHash.
 	dimHash uint64
-	// err is an error that occured during construction. It is reported on
+	// err is an error that occurred during construction. It is reported on
 	// registration time.
 	err error
 }
@@ -103,7 +98,7 @@ func NewDesc(fqName, help string, variableLabels []string, constLabels Labels) *
 		d.err = errors.New("empty help string")
 		return d
 	}
-	if !metricNameRE.MatchString(fqName) {
+	if !model.IsValidMetricName(model.LabelValue(fqName)) {
 		d.err = fmt.Errorf("%q is not a valid metric name", fqName)
 		return d
 	}
@@ -200,6 +195,6 @@ func (d *Desc) String() string {
 }
 
 func checkLabelName(l string) bool {
-	return labelNameRE.MatchString(l) &&
+	return model.LabelName(l).IsValid() &&
 		!strings.HasPrefix(l, reservedLabelPrefix)
 }

vendor/github.com/prometheus/client_golang/prometheus/doc.go

@@ -17,7 +17,7 @@
 // Pushgateway (package push).
 //
 // All exported functions and methods are safe to be used concurrently unless
-//specified otherwise.
+// specified otherwise.
 //
 // A Basic Example
 //
@@ -59,7 +59,7 @@
 //    	// The Handler function provides a default handler to expose metrics
 //    	// via an HTTP server. "/metrics" is the usual endpoint for that.
 //    	http.Handle("/metrics", promhttp.Handler())
-//    	http.ListenAndServe(":8080", nil)
+//      log.Fatal(http.ListenAndServe(":8080", nil))
 //    }
 //
 //
@@ -69,7 +69,7 @@
 // Metrics
 //
 // The number of exported identifiers in this package might appear a bit
-// overwhelming. Hovever, in addition to the basic plumbing shown in the example
+// overwhelming. However, in addition to the basic plumbing shown in the example
 // above, you only need to understand the different metric types and their
 // vector versions for basic usage.
 //
@@ -95,8 +95,8 @@
 // SummaryVec, HistogramVec, and UntypedVec are not.
 //
 // To create instances of Metrics and their vector versions, you need a suitable
-// …Opts struct, i.e. GaugeOpts, CounterOpts, SummaryOpts,
+// …Opts struct, i.e. GaugeOpts, CounterOpts, SummaryOpts, HistogramOpts, or
-// HistogramOpts, or UntypedOpts.
+// UntypedOpts.
 //
 // Custom Collectors and constant Metrics
 //
@@ -114,8 +114,8 @@
 // Metric instances “on the fly” using NewConstMetric, NewConstHistogram, and
 // NewConstSummary (and their respective Must… versions). That will happen in
 // the Collect method. The Describe method has to return separate Desc
-// instances, representative of the “throw-away” metrics to be created
+// instances, representative of the “throw-away” metrics to be created later.
-// later. NewDesc comes in handy to create those Desc instances.
+// NewDesc comes in handy to create those Desc instances.
 //
 // The Collector example illustrates the use case. You can also look at the
 // source code of the processCollector (mirroring process metrics), the
@@ -129,32 +129,32 @@
 // Advanced Uses of the Registry
 //
 // While MustRegister is the by far most common way of registering a Collector,
-// sometimes you might want to handle the errors the registration might
+// sometimes you might want to handle the errors the registration might cause.
-// cause. As suggested by the name, MustRegister panics if an error occurs. With
+// As suggested by the name, MustRegister panics if an error occurs. With the
-// the Register function, the error is returned and can be handled.
+// Register function, the error is returned and can be handled.
 //
 // An error is returned if the registered Collector is incompatible or
 // inconsistent with already registered metrics. The registry aims for
-// consistency of the collected metrics according to the Prometheus data
+// consistency of the collected metrics according to the Prometheus data model.
-// model. Inconsistencies are ideally detected at registration time, not at
+// Inconsistencies are ideally detected at registration time, not at collect
-// collect time. The former will usually be detected at start-up time of a
+// time. The former will usually be detected at start-up time of a program,
-// program, while the latter will only happen at scrape time, possibly not even
+// while the latter will only happen at scrape time, possibly not even on the
-// on the first scrape if the inconsistency only becomes relevant later. That is
+// first scrape if the inconsistency only becomes relevant later. That is the
-// the main reason why a Collector and a Metric have to describe themselves to
+// main reason why a Collector and a Metric have to describe themselves to the
-// the registry.
+// registry.
 //
 // So far, everything we did operated on the so-called default registry, as it
 // can be found in the global DefaultRegistry variable. With NewRegistry, you
 // can create a custom registry, or you can even implement the Registerer or
-// Gatherer interfaces yourself. The methods Register and Unregister work in
+// Gatherer interfaces yourself. The methods Register and Unregister work in the
-// the same way on a custom registry as the global functions Register and
+// same way on a custom registry as the global functions Register and Unregister
-// Unregister on the default registry.
+// on the default registry.
 //
-// There are a number of uses for custom registries: You can use registries
+// There are a number of uses for custom registries: You can use registries with
-// with special properties, see NewPedanticRegistry. You can avoid global state,
+// special properties, see NewPedanticRegistry. You can avoid global state, as
-// as it is imposed by the DefaultRegistry. You can use multiple registries at
+// it is imposed by the DefaultRegistry. You can use multiple registries at the
-// the same time to expose different metrics in different ways. You can use
+// same time to expose different metrics in different ways. You can use separate
-// separate registries for testing purposes.
+// registries for testing purposes.
 //
 // Also note that the DefaultRegistry comes registered with a Collector for Go
 // runtime metrics (via NewGoCollector) and a Collector for process metrics (via
@@ -166,16 +166,20 @@
 // The Registry implements the Gatherer interface. The caller of the Gather
 // method can then expose the gathered metrics in some way. Usually, the metrics
 // are served via HTTP on the /metrics endpoint. That's happening in the example
-// above. The tools to expose metrics via HTTP are in the promhttp
+// above. The tools to expose metrics via HTTP are in the promhttp sub-package.
-// sub-package. (The top-level functions in the prometheus package are
+// (The top-level functions in the prometheus package are deprecated.)
-// deprecated.)
 //
 // Pushing to the Pushgateway
 //
 // Function for pushing to the Pushgateway can be found in the push sub-package.
 //
+// Graphite Bridge
+//
+// Functions and examples to push metrics from a Gatherer to Graphite can be
+// found in the graphite sub-package.
+//
 // Other Means of Exposition
 //
-// More ways of exposing metrics can easily be added. Sending metrics to
+// More ways of exposing metrics can easily be added by following the approaches
-// Graphite would be an example that will soon be implemented.
+// of the existing implementations.
 package prometheus

vendor/github.com/prometheus/client_golang/prometheus/gauge.go

@@ -27,16 +27,21 @@ type Gauge interface {
 
 	// Set sets the Gauge to an arbitrary value.
 	Set(float64)
-	// Inc increments the Gauge by 1.
+	// Inc increments the Gauge by 1. Use Add to increment it by arbitrary
+	// values.
 	Inc()
-	// Dec decrements the Gauge by 1.
+	// Dec decrements the Gauge by 1. Use Sub to decrement it by arbitrary
+	// values.
 	Dec()
-	// Add adds the given value to the Gauge. (The value can be
+	// Add adds the given value to the Gauge. (The value can be negative,
-	// negative, resulting in a decrease of the Gauge.)
+	// resulting in a decrease of the Gauge.)
 	Add(float64)
 	// Sub subtracts the given value from the Gauge. (The value can be
 	// negative, resulting in an increase of the Gauge.)
 	Sub(float64)
+
+	// SetToCurrentTime sets the Gauge to the current Unix time in seconds.
+	SetToCurrentTime()
 }
 
 // GaugeOpts is an alias for Opts. See there for doc comments.

vendor/github.com/prometheus/client_golang/prometheus/go_collector.go

@@ -8,8 +8,9 @@ import (
 )
 
 type goCollector struct {
-	goroutines Gauge
+	goroutinesDesc *Desc
-	gcDesc     *Desc
+	threadsDesc    *Desc
+	gcDesc         *Desc
 
 	// metrics to describe and collect
 	metrics memStatsMetrics
@@ -19,11 +20,14 @@ type goCollector struct {
 // go process.
 func NewGoCollector() Collector {
 	return &goCollector{
-		goroutines: NewGauge(GaugeOpts{
+		goroutinesDesc: NewDesc(
-			Namespace: "go",
+			"go_goroutines",
-			Name:      "goroutines",
+			"Number of goroutines that currently exist.",
-			Help:      "Number of goroutines that currently exist.",
+			nil, nil),
-		}),
+		threadsDesc: NewDesc(
+			"go_threads",
+			"Number of OS threads created",
+			nil, nil),
 		gcDesc: NewDesc(
 			"go_gc_duration_seconds",
 			"A summary of the GC invocation durations.",
@@ -48,7 +52,7 @@ func NewGoCollector() Collector {
 			}, {
 				desc: NewDesc(
 					memstatNamespace("sys_bytes"),
-					"Number of bytes obtained by system. Sum of all system allocations.",
+					"Number of bytes obtained from system.",
 					nil, nil,
 				),
 				eval:    func(ms *runtime.MemStats) float64 { return float64(ms.Sys) },
@@ -111,12 +115,12 @@ func NewGoCollector() Collector {
 				valType: GaugeValue,
 			}, {
 				desc: NewDesc(
-					memstatNamespace("heap_released_bytes_total"),
+					memstatNamespace("heap_released_bytes"),
-					"Total number of heap bytes released to OS.",
+					"Number of heap bytes released to OS.",
 					nil, nil,
 				),
 				eval:    func(ms *runtime.MemStats) float64 { return float64(ms.HeapReleased) },
-				valType: CounterValue,
+				valType: GaugeValue,
 			}, {
 				desc: NewDesc(
 					memstatNamespace("heap_objects"),
@@ -213,6 +217,14 @@ func NewGoCollector() Collector {
 				),
 				eval:    func(ms *runtime.MemStats) float64 { return float64(ms.LastGC) / 1e9 },
 				valType: GaugeValue,
+			}, {
+				desc: NewDesc(
+					memstatNamespace("gc_cpu_fraction"),
+					"The fraction of this program's available CPU time used by the GC since the program started.",
+					nil, nil,
+				),
+				eval:    func(ms *runtime.MemStats) float64 { return ms.GCCPUFraction },
+				valType: GaugeValue,
 			},
 		},
 	}
@@ -224,9 +236,9 @@ func memstatNamespace(s string) string {
 
 // Describe returns all descriptions of the collector.
 func (c *goCollector) Describe(ch chan<- *Desc) {
-	ch <- c.goroutines.Desc()
+	ch <- c.goroutinesDesc
+	ch <- c.threadsDesc
 	ch <- c.gcDesc
-
 	for _, i := range c.metrics {
 		ch <- i.desc
 	}
@@ -234,8 +246,9 @@ func (c *goCollector) Describe(ch chan<- *Desc) {
 
 // Collect returns the current state of all metrics of the collector.
 func (c *goCollector) Collect(ch chan<- Metric) {
-	c.goroutines.Set(float64(runtime.NumGoroutine()))
+	ch <- MustNewConstMetric(c.goroutinesDesc, GaugeValue, float64(runtime.NumGoroutine()))
-	ch <- c.goroutines
+	n, _ := runtime.ThreadCreateProfile(nil)
+	ch <- MustNewConstMetric(c.threadsDesc, GaugeValue, float64(n))
 
 	var stats debug.GCStats
 	stats.PauseQuantiles = make([]time.Duration, 5)

vendor/github.com/prometheus/client_golang/prometheus/http.go

@@ -62,7 +62,7 @@ func giveBuf(buf *bytes.Buffer) {
 //
 // Deprecated: Please note the issues described in the doc comment of
 // InstrumentHandler. You might want to consider using promhttp.Handler instead
-// (which is non instrumented).
+// (which is not instrumented).
 func Handler() http.Handler {
 	return InstrumentHandler("prometheus", UninstrumentedHandler())
 }
@@ -172,6 +172,9 @@ func nowSeries(t ...time.Time) nower {
 // httputil.ReverseProxy is a prominent example for a handler
 // performing such writes.
 //
+// - It has additional issues with HTTP/2, cf.
+// https://github.com/prometheus/client_golang/issues/272.
+//
 // Upcoming versions of this package will provide ways of instrumenting HTTP
 // handlers that are more flexible and have fewer issues. Please prefer direct
 // instrumentation in the meantime.
@@ -190,6 +193,7 @@ func InstrumentHandlerFunc(handlerName string, handlerFunc func(http.ResponseWri
 		SummaryOpts{
 			Subsystem:   "http",
 			ConstLabels: Labels{"handler": handlerName},
+			Objectives:  map[float64]float64{0.5: 0.05, 0.9: 0.01, 0.99: 0.001},
 		},
 		handlerFunc,
 	)
@@ -245,34 +249,52 @@ func InstrumentHandlerFuncWithOpts(opts SummaryOpts, handlerFunc func(http.Respo
 		},
 		instLabels,
 	)
+	if err := Register(reqCnt); err != nil {
+		if are, ok := err.(AlreadyRegisteredError); ok {
+			reqCnt = are.ExistingCollector.(*CounterVec)
+		} else {
+			panic(err)
+		}
+	}
 
 	opts.Name = "request_duration_microseconds"
 	opts.Help = "The HTTP request latencies in microseconds."
 	reqDur := NewSummary(opts)
+	if err := Register(reqDur); err != nil {
+		if are, ok := err.(AlreadyRegisteredError); ok {
+			reqDur = are.ExistingCollector.(Summary)
+		} else {
+			panic(err)
+		}
+	}
 
 	opts.Name = "request_size_bytes"
 	opts.Help = "The HTTP request sizes in bytes."
 	reqSz := NewSummary(opts)
+	if err := Register(reqSz); err != nil {
+		if are, ok := err.(AlreadyRegisteredError); ok {
+			reqSz = are.ExistingCollector.(Summary)
+		} else {
+			panic(err)
+		}
+	}
 
 	opts.Name = "response_size_bytes"
 	opts.Help = "The HTTP response sizes in bytes."
 	resSz := NewSummary(opts)
-
+	if err := Register(resSz); err != nil {
-	regReqCnt := MustRegisterOrGet(reqCnt).(*CounterVec)
+		if are, ok := err.(AlreadyRegisteredError); ok {
-	regReqDur := MustRegisterOrGet(reqDur).(Summary)
+			resSz = are.ExistingCollector.(Summary)
-	regReqSz := MustRegisterOrGet(reqSz).(Summary)
+		} else {
-	regResSz := MustRegisterOrGet(resSz).(Summary)
+			panic(err)
+		}
+	}
 
 	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		now := time.Now()
 
 		delegate := &responseWriterDelegator{ResponseWriter: w}
-		out := make(chan int)
+		out := computeApproximateRequestSize(r)
-		urlLen := 0
-		if r.URL != nil {
-			urlLen = len(r.URL.String())
-		}
-		go computeApproximateRequestSize(r, out, urlLen)
 
 		_, cn := w.(http.CloseNotifier)
 		_, fl := w.(http.Flusher)
@@ -290,30 +312,44 @@ func InstrumentHandlerFuncWithOpts(opts SummaryOpts, handlerFunc func(http.Respo
 
 		method := sanitizeMethod(r.Method)
 		code := sanitizeCode(delegate.status)
-		regReqCnt.WithLabelValues(method, code).Inc()
+		reqCnt.WithLabelValues(method, code).Inc()
-		regReqDur.Observe(elapsed)
+		reqDur.Observe(elapsed)
-		regResSz.Observe(float64(delegate.written))
+		resSz.Observe(float64(delegate.written))
-		regReqSz.Observe(float64(<-out))
+		reqSz.Observe(float64(<-out))
 	})
 }
 
-func computeApproximateRequestSize(r *http.Request, out chan int, s int) {
+func computeApproximateRequestSize(r *http.Request) <-chan int {
-	s += len(r.Method)
+	// Get URL length in current go routine for avoiding a race condition.
-	s += len(r.Proto)
+	// HandlerFunc that runs in parallel may modify the URL.
-	for name, values := range r.Header {
+	s := 0
-		s += len(name)
+	if r.URL != nil {
-		for _, value := range values {
+		s += len(r.URL.String())
-			s += len(value)
+	}
+
+	out := make(chan int, 1)
+
+	go func() {
+		s += len(r.Method)
+		s += len(r.Proto)
+		for name, values := range r.Header {
+			s += len(name)
+			for _, value := range values {
+				s += len(value)
+			}
 		}
-	}
+		s += len(r.Host)
-	s += len(r.Host)
 
-	// N.B. r.Form and r.MultipartForm are assumed to be included in r.URL.
+		// N.B. r.Form and r.MultipartForm are assumed to be included in r.URL.
 
-	if r.ContentLength != -1 {
+		if r.ContentLength != -1 {
-		s += int(r.ContentLength)
+			s += int(r.ContentLength)
-	}
+		}
-	out <- s
+		out <- s
+		close(out)
+	}()
+
+	return out
 }
 
 type responseWriterDelegator struct {

vendor/github.com/prometheus/client_golang/prometheus/process_collector.go

@@ -19,10 +19,10 @@ type processCollector struct {
 	pid             int
 	collectFn       func(chan<- Metric)
 	pidFn           func() (int, error)
-	cpuTotal        Counter
+	cpuTotal        *Desc
-	openFDs, maxFDs Gauge
+	openFDs, maxFDs *Desc
-	vsize, rss      Gauge
+	vsize, rss      *Desc
-	startTime       Gauge
+	startTime       *Desc
 }
 
 // NewProcessCollector returns a collector which exports the current state of
@@ -44,40 +44,45 @@ func NewProcessCollectorPIDFn(
 	pidFn func() (int, error),
 	namespace string,
 ) Collector {
+	ns := ""
+	if len(namespace) > 0 {
+		ns = namespace + "_"
+	}
+
 	c := processCollector{
 		pidFn:     pidFn,
 		collectFn: func(chan<- Metric) {},
 
-		cpuTotal: NewCounter(CounterOpts{
+		cpuTotal: NewDesc(
-			Namespace: namespace,
+			ns+"process_cpu_seconds_total",
-			Name:      "process_cpu_seconds_total",
+			"Total user and system CPU time spent in seconds.",
-			Help:      "Total user and system CPU time spent in seconds.",
+			nil, nil,
-		}),
+		),
-		openFDs: NewGauge(GaugeOpts{
+		openFDs: NewDesc(
-			Namespace: namespace,
+			ns+"process_open_fds",
-			Name:      "process_open_fds",
+			"Number of open file descriptors.",
-			Help:      "Number of open file descriptors.",
+			nil, nil,
-		}),
+		),
-		maxFDs: NewGauge(GaugeOpts{
+		maxFDs: NewDesc(
-			Namespace: namespace,
+			ns+"process_max_fds",
-			Name:      "process_max_fds",
+			"Maximum number of open file descriptors.",
-			Help:      "Maximum number of open file descriptors.",
+			nil, nil,
-		}),
+		),
-		vsize: NewGauge(GaugeOpts{
+		vsize: NewDesc(
-			Namespace: namespace,
+			ns+"process_virtual_memory_bytes",
-			Name:      "process_virtual_memory_bytes",
+			"Virtual memory size in bytes.",
-			Help:      "Virtual memory size in bytes.",
+			nil, nil,
-		}),
+		),
-		rss: NewGauge(GaugeOpts{
+		rss: NewDesc(
-			Namespace: namespace,
+			ns+"process_resident_memory_bytes",
-			Name:      "process_resident_memory_bytes",
+			"Resident memory size in bytes.",
-			Help:      "Resident memory size in bytes.",
+			nil, nil,
-		}),
+		),
-		startTime: NewGauge(GaugeOpts{
+		startTime: NewDesc(
-			Namespace: namespace,
+			ns+"process_start_time_seconds",
-			Name:      "process_start_time_seconds",
+			"Start time of the process since unix epoch in seconds.",
-			Help:      "Start time of the process since unix epoch in seconds.",
+			nil, nil,
-		}),
+		),
 	}
 
 	// Set up process metric collection if supported by the runtime.
@@ -90,12 +95,12 @@ func NewProcessCollectorPIDFn(
 
 // Describe returns all descriptions of the collector.
 func (c *processCollector) Describe(ch chan<- *Desc) {
-	ch <- c.cpuTotal.Desc()
+	ch <- c.cpuTotal
-	ch <- c.openFDs.Desc()
+	ch <- c.openFDs
-	ch <- c.maxFDs.Desc()
+	ch <- c.maxFDs
-	ch <- c.vsize.Desc()
+	ch <- c.vsize
-	ch <- c.rss.Desc()
+	ch <- c.rss
-	ch <- c.startTime.Desc()
+	ch <- c.startTime
 }
 
 // Collect returns the current state of all metrics of the collector.
@@ -117,26 +122,19 @@ func (c *processCollector) processCollect(ch chan<- Metric) {
 	}
 
 	if stat, err := p.NewStat(); err == nil {
-		c.cpuTotal.Set(stat.CPUTime())
+		ch <- MustNewConstMetric(c.cpuTotal, CounterValue, stat.CPUTime())
-		ch <- c.cpuTotal
+		ch <- MustNewConstMetric(c.vsize, GaugeValue, float64(stat.VirtualMemory()))
-		c.vsize.Set(float64(stat.VirtualMemory()))
+		ch <- MustNewConstMetric(c.rss, GaugeValue, float64(stat.ResidentMemory()))
-		ch <- c.vsize
-		c.rss.Set(float64(stat.ResidentMemory()))
-		ch <- c.rss
-
 		if startTime, err := stat.StartTime(); err == nil {
-			c.startTime.Set(startTime)
+			ch <- MustNewConstMetric(c.startTime, GaugeValue, startTime)
-			ch <- c.startTime
 		}
 	}
 
 	if fds, err := p.FileDescriptorsLen(); err == nil {
-		c.openFDs.Set(float64(fds))
+		ch <- MustNewConstMetric(c.openFDs, GaugeValue, float64(fds))
-		ch <- c.openFDs
 	}
 
 	if limits, err := p.NewLimits(); err == nil {
-		c.maxFDs.Set(float64(limits.OpenFiles))
+		ch <- MustNewConstMetric(c.maxFDs, GaugeValue, float64(limits.OpenFiles))
-		ch <- c.maxFDs
 	}
 }

vendor/github.com/prometheus/client_golang/prometheus/registry.go

@@ -152,38 +152,6 @@ func MustRegister(cs ...Collector) {
 	DefaultRegisterer.MustRegister(cs...)
 }
 
-// RegisterOrGet registers the provided Collector with the DefaultRegisterer and
-// returns the Collector, unless an equal Collector was registered before, in
-// which case that Collector is returned.
-//
-// Deprecated: RegisterOrGet is merely a convenience function for the
-// implementation as described in the documentation for
-// AlreadyRegisteredError. As the use case is relatively rare, this function
-// will be removed in a future version of this package to clean up the
-// namespace.
-func RegisterOrGet(c Collector) (Collector, error) {
-	if err := Register(c); err != nil {
-		if are, ok := err.(AlreadyRegisteredError); ok {
-			return are.ExistingCollector, nil
-		}
-		return nil, err
-	}
-	return c, nil
-}
-
-// MustRegisterOrGet behaves like RegisterOrGet but panics instead of returning
-// an error.
-//
-// Deprecated: This is deprecated for the same reason RegisterOrGet is. See
-// there for details.
-func MustRegisterOrGet(c Collector) Collector {
-	c, err := RegisterOrGet(c)
-	if err != nil {
-		panic(err)
-	}
-	return c
-}
-
 // Unregister removes the registration of the provided Collector from the
 // DefaultRegisterer.
 //
@@ -201,25 +169,6 @@ func (gf GathererFunc) Gather() ([]*dto.MetricFamily, error) {
 	return gf()
 }
 
-// SetMetricFamilyInjectionHook replaces the DefaultGatherer with one that
-// gathers from the previous DefaultGatherers but then merges the MetricFamily
-// protobufs returned from the provided hook function with the MetricFamily
-// protobufs returned from the original DefaultGatherer.
-//
-// Deprecated: This function manipulates the DefaultGatherer variable. Consider
-// the implications, i.e. don't do this concurrently with any uses of the
-// DefaultGatherer. In the rare cases where you need to inject MetricFamily
-// protobufs directly, it is recommended to use a custom Registry and combine it
-// with a custom Gatherer using the Gatherers type (see
-// there). SetMetricFamilyInjectionHook only exists for compatibility reasons
-// with previous versions of this package.
-func SetMetricFamilyInjectionHook(hook func() []*dto.MetricFamily) {
-	DefaultGatherer = Gatherers{
-		DefaultGatherer,
-		GathererFunc(func() ([]*dto.MetricFamily, error) { return hook(), nil }),
-	}
-}
-
 // AlreadyRegisteredError is returned by the Register method if the Collector to
 // be registered has already been registered before, or a different Collector
 // that collects the same metrics has been registered before. Registration fails
@@ -294,7 +243,7 @@ func (r *Registry) Register(c Collector) error {
 	}()
 	r.mtx.Lock()
 	defer r.mtx.Unlock()
-	// Coduct various tests...
+	// Conduct various tests...
 	for desc := range descChan {
 
 		// Is the descriptor valid at all?
@@ -447,7 +396,7 @@ func (r *Registry) Gather() ([]*dto.MetricFamily, error) {
 
 	// Drain metricChan in case of premature return.
 	defer func() {
-		for _ = range metricChan {
+		for range metricChan {
 		}
 	}()
 
@@ -683,7 +632,7 @@ func (s metricSorter) Less(i, j int) bool {
 	return s[i].GetTimestampMs() < s[j].GetTimestampMs()
 }
 
-// normalizeMetricFamilies returns a MetricFamily slice whith empty
+// normalizeMetricFamilies returns a MetricFamily slice with empty
 // MetricFamilies pruned and the remaining MetricFamilies sorted by name within
 // the slice, with the contained Metrics sorted within each MetricFamily.
 func normalizeMetricFamilies(metricFamiliesByName map[string]*dto.MetricFamily) []*dto.MetricFamily {

vendor/github.com/prometheus/client_golang/prometheus/summary.go

@@ -54,6 +54,9 @@ type Summary interface {
 }
 
 // DefObjectives are the default Summary quantile values.
+//
+// Deprecated: DefObjectives will not be used as the default objectives in
+// v0.10 of the library. The default Summary will have no quantiles then.
 var (
 	DefObjectives = map[float64]float64{0.5: 0.05, 0.9: 0.01, 0.99: 0.001}
 
@@ -113,9 +116,15 @@ type SummaryOpts struct {
 	ConstLabels Labels
 
 	// Objectives defines the quantile rank estimates with their respective
-	// absolute error. If Objectives[q] = e, then the value reported
+	// absolute error. If Objectives[q] = e, then the value reported for q
-	// for q will be the φ-quantile value for some φ between q-e and q+e.
+	// will be the φ-quantile value for some φ between q-e and q+e.  The
-	// The default value is DefObjectives.
+	// default value is DefObjectives. It is used if Objectives is left at
+	// its zero value (i.e. nil). To create a Summary without Objectives,
+	// set it to an empty map (i.e. map[float64]float64{}).
+	//
+	// Deprecated: Note that the current value of DefObjectives is
+	// deprecated. It will be replaced by an empty map in v0.10 of the
+	// library. Please explicitly set Objectives to the desired value.
 	Objectives map[float64]float64
 
 	// MaxAge defines the duration for which an observation stays relevant
@@ -183,7 +192,7 @@ func newSummary(desc *Desc, opts SummaryOpts, labelValues ...string) Summary {
 		}
 	}
 
-	if len(opts.Objectives) == 0 {
+	if opts.Objectives == nil {
 		opts.Objectives = DefObjectives
 	}
 

vendor/github.com/prometheus/client_golang/prometheus/timer.go

@@ -0,0 +1,74 @@
+// Copyright 2016 The Prometheus Authors
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package prometheus
+
+import "time"
+
+// Observer is the interface that wraps the Observe method, which is used by
+// Histogram and Summary to add observations.
+type Observer interface {
+	Observe(float64)
+}
+
+// The ObserverFunc type is an adapter to allow the use of ordinary
+// functions as Observers. If f is a function with the appropriate
+// signature, ObserverFunc(f) is an Observer that calls f.
+//
+// This adapter is usually used in connection with the Timer type, and there are
+// two general use cases:
+//
+// The most common one is to use a Gauge as the Observer for a Timer.
+// See the "Gauge" Timer example.
+//
+// The more advanced use case is to create a function that dynamically decides
+// which Observer to use for observing the duration. See the "Complex" Timer
+// example.
+type ObserverFunc func(float64)
+
+// Observe calls f(value). It implements Observer.
+func (f ObserverFunc) Observe(value float64) {
+	f(value)
+}
+
+// Timer is a helper type to time functions. Use NewTimer to create new
+// instances.
+type Timer struct {
+	begin    time.Time
+	observer Observer
+}
+
+// NewTimer creates a new Timer. The provided Observer is used to observe a
+// duration in seconds. Timer is usually used to time a function call in the
+// following way:
+//    func TimeMe() {
+//        timer := NewTimer(myHistogram)
+//        defer timer.ObserveDuration()
+//        // Do actual work.
+//    }
+func NewTimer(o Observer) *Timer {
+	return &Timer{
+		begin:    time.Now(),
+		observer: o,
+	}
+}
+
+// ObserveDuration records the duration passed since the Timer was created with
+// NewTimer. It calls the Observe method of the Observer provided during
+// construction with the duration in seconds as an argument. ObserveDuration is
+// usually called with a defer statement.
+func (t *Timer) ObserveDuration() {
+	if t.observer != nil {
+		t.observer.Observe(time.Since(t.begin).Seconds())
+	}
+}

vendor/github.com/prometheus/client_golang/prometheus/untyped.go

@@ -20,6 +20,11 @@ package prometheus
 // no type information is implied.
 //
 // To create Untyped instances, use NewUntyped.
+//
+// Deprecated: The Untyped type is deprecated because it doesn't make sense in
+// direct instrumentation. If you need to mirror an external metric of unknown
+// type (usually while writing exporters), Use MustNewConstMetric to create an
+// untyped metric instance on the fly.
 type Untyped interface {
 	Metric
 	Collector

vendor/github.com/prometheus/client_golang/prometheus/value.go

@@ -19,6 +19,7 @@ import (
 	"math"
 	"sort"
 	"sync/atomic"
+	"time"
 
 	dto "github.com/prometheus/client_model/go"
 
@@ -43,7 +44,7 @@ var errInconsistentCardinality = errors.New("inconsistent label cardinality")
 // ValueType. This is a low-level building block used by the library to back the
 // implementations of Counter, Gauge, and Untyped.
 type value struct {
-	// valBits containst the bits of the represented float64 value. It has
+	// valBits contains the bits of the represented float64 value. It has
 	// to go first in the struct to guarantee alignment for atomic
 	// operations.  http://golang.org/pkg/sync/atomic/#pkg-note-BUG
 	valBits uint64
@@ -80,6 +81,10 @@ func (v *value) Set(val float64) {
 	atomic.StoreUint64(&v.valBits, math.Float64bits(val))
 }
 
+func (v *value) SetToCurrentTime() {
+	v.Set(float64(time.Now().UnixNano()) / 1e9)
+}
+
 func (v *value) Inc() {
 	v.Add(1)
 }

vendor/vendor.json

@@ -603,10 +603,10 @@
 			"revisionTime": "2016-06-15T09:26:46Z"
 		},
 		{
-			"checksumSHA1": "KkB+77Ziom7N6RzSbyUwYGrmDeU=",
+			"checksumSHA1": "d2irkxoHgazkTuLIvJGiYwagl8o=",
 			"path": "github.com/prometheus/client_golang/prometheus",
-			"revision": "c5b7fccd204277076155f10851dad72b76a49317",
+			"revision": "08fd2e12372a66e68e30523c7642e0cbc3e4fbde",
-			"revisionTime": "2016-08-17T15:48:24Z"
+			"revisionTime": "2017-04-01T10:34:46Z"
 		},
 		{
 			"checksumSHA1": "DvwvOlPNAgRntBzt3b3OSRMS2N4=",
@@ -615,10 +615,10 @@
 			"revisionTime": "2015-02-12T10:17:44Z"
 		},
 		{
-			"checksumSHA1": "jG8qYuDUuaZeflt4JxBBdyQBsXw=",
+			"checksumSHA1": "Wtpzndm/+bdwwNU5PCTfb4oUhc8=",
 			"path": "github.com/prometheus/common/expfmt",
-			"revision": "dd2f054febf4a6c00f2343686efb775948a8bff4",
+			"revision": "49fee292b27bfff7f354ee0f64e1bc4850462edf",
-			"revisionTime": "2017-01-08T23:12:12Z"
+			"revisionTime": "2017-02-20T10:38:46Z"
 		},
 		{
 			"checksumSHA1": "GWlM3d2vPYyNATtTFgftS10/A9w=",
@@ -633,10 +633,10 @@
 			"revisionTime": "2017-01-08T23:12:12Z"
 		},
 		{
-			"checksumSHA1": "vopCLXHzYm+3l5fPKOf4/fQwrCM=",
+			"checksumSHA1": "0LL9u9tfv1KPBjNEiMDP6q7lpog=",
 			"path": "github.com/prometheus/common/model",
-			"revision": "3007b6072c17c8d985734e6e19b1dea9174e13d3",
+			"revision": "49fee292b27bfff7f354ee0f64e1bc4850462edf",
-			"revisionTime": "2017-02-19T00:35:58+01:00"
+			"revisionTime": "2017-02-20T10:38:46Z"
 		},
 		{
 			"checksumSHA1": "ZbbESWBHHcPUJ/A5yrzKhTHuPc8=",

web/api/v1/api.go

@@ -19,6 +19,7 @@ import (
 	"fmt"
 	"math"
 	"net/http"
+	"net/url"
 	"strconv"
 	"time"
 
@@ -50,6 +51,7 @@ const (
 	errorCanceled           = "canceled"
 	errorExec               = "execution"
 	errorBadData            = "bad_data"
+	errorInternal           = "internal"
 )
 
 var corsHeaders = map[string]string{
@@ -73,7 +75,7 @@ type targetRetriever interface {
 }
 
 type alertmanagerRetriever interface {
-	Alertmanagers() []string
+	Alertmanagers() []*url.URL
 }
 
 type response struct {
@@ -194,6 +196,8 @@ func (api *API) query(r *http.Request) (interface{}, *apiError) {
 			return nil, &apiError{errorCanceled, res.Err}
 		case promql.ErrQueryTimeout:
 			return nil, &apiError{errorTimeout, res.Err}
+		case promql.ErrStorage:
+			return nil, &apiError{errorInternal, res.Err}
 		}
 		return nil, &apiError{errorExec, res.Err}
 	}
@@ -362,7 +366,7 @@ func (api *API) dropSeries(r *http.Request) (interface{}, *apiError) {
 	}
 
 	// TODO(fabxc): temporarily disabled
-	panic("disabled")
+	return nil, &apiError{errorExec, fmt.Errorf("temporarily disabled")}
 
 	// numDeleted := 0
 	// for _, s := range r.Form["match[]"] {
@@ -442,8 +446,8 @@ func (api *API) alertmanagers(r *http.Request) (interface{}, *apiError) {
 	urls := api.alertmanagerRetriever.Alertmanagers()
 	ams := &AlertmanagerDiscovery{ActiveAlertmanagers: make([]*AlertmanagerTarget, len(urls))}
 
-	for i := range urls {
+	for i, url := range urls {
-		ams.ActiveAlertmanagers[i] = &AlertmanagerTarget{URL: urls[i]}
+		ams.ActiveAlertmanagers[i] = &AlertmanagerTarget{URL: url.String()}
 	}
 
 	return ams, nil
@@ -474,6 +478,8 @@ func respondError(w http.ResponseWriter, apiErr *apiError, data interface{}) {
 		code = 422
 	case errorCanceled, errorTimeout:
 		code = http.StatusServiceUnavailable
+	case errorInternal:
+		code = http.StatusInternalServerError
 	default:
 		code = http.StatusInternalServerError
 	}

web/api/v1/api_test.go

@@ -41,9 +41,9 @@ func (f targetRetrieverFunc) Targets() []*retrieval.Target {
 	return f()
 }
 
-type alertmanagerRetrieverFunc func() []string
+type alertmanagerRetrieverFunc func() []*url.URL
 
-func (f alertmanagerRetrieverFunc) Alertmanagers() []string {
+func (f alertmanagerRetrieverFunc) Alertmanagers() []*url.URL {
 	return f()
 }
 
@@ -79,8 +79,12 @@ func TestEndpoints(t *testing.T) {
 		}
 	})
 
-	ar := alertmanagerRetrieverFunc(func() []string {
+	ar := alertmanagerRetrieverFunc(func() []*url.URL {
-		return []string{"http://alertmanager.example.com:8080/api/v1/alerts"}
+		return []*url.URL{{
+			Scheme: "http",
+			Host:   "alertmanager.example.com:8080",
+			Path:   "/api/v1/alerts",
+		}}
 	})
 
 	api := &API{
@@ -430,15 +434,27 @@ func TestEndpoints(t *testing.T) {
 		// 	}{2},
 		// }, {
 		// 	endpoint: api.targets,
-		// 	response: []*Target{
+		// 	response: &TargetDiscovery{
-		// 		&Target{
+		// 		ActiveTargets: []*Target{
-		// 			DiscoveredLabels: nil,
+		// 			{
-		// 			Labels:           nil,
+		// 				DiscoveredLabels: model.LabelSet{},
-		// 			ScrapeUrl:        "http://example.com:8080/metrics",
+		// 				Labels:           model.LabelSet{},
-		// 			Health:           "unknown",
+		// 				ScrapeURL:        "http://example.com:8080/metrics",
+		// 				Health:           "unknown",
+		// 			},
 		// 		},
 		// 	},
 		// },
+		{
+			endpoint: api.alertmanagers,
+			response: &AlertmanagerDiscovery{
+				ActiveAlertmanagers: []*AlertmanagerTarget{
+					{
+						URL: "http://alertmanager.example.com:8080/api/v1/alerts",
+					},
+				},
+			},
+		},
 	}
 
 	for _, test := range tests {

web/ui/static/css/graph.css

@@ -44,23 +44,27 @@ body {
 .legend {
   display: inline-block;
   vertical-align: top;
-  margin: 0 0 0 40px;
+  margin: 0 0 0 60px;
 }
 
 .graph_area {
   position: relative;
   font-family: Arial, Helvetica, sans-serif;
-  margin: 5px 0 5px 0;
+  margin: 5px 0 5px 20px;
 }
 
 .y_axis {
-  overflow: hidden;
+  overflow: visible;
   position: absolute;
   top: 1px;
   bottom: 0;
   width: 40px;
 }
 
+.y_axis svg {
+  overflow: visible;
+}
+
 .graph .detail .item.active {
   line-height: 1.4em;
   padding: 0.5em;
@@ -126,7 +130,7 @@ input[name="end_input"], input[name="range_input"] {
 }
 
 .prometheus_input_group.range_input {
-  margin-left: 39px;
+  margin-left: 59px;
 }
 
 .prometheus_input_group .btn {

web/ui/static/js/graph.js

@@ -415,7 +415,7 @@ Prometheus.Graph.prototype.submitQuery = function() {
           return;
         }
         var duration = new Date().getTime() - startTime;
-        var totalTimeSeries = xhr.responseJSON.data.result.length;
+        var totalTimeSeries = (xhr.responseJSON.data !== undefined) ? xhr.responseJSON.data.result.length : 0;
         self.evalStats.html("Load time: " + duration + "ms <br /> Resolution: " + resolution + "s <br />" + "Total time series: " + totalTimeSeries);
         self.spinner.hide();
       }
@@ -556,6 +556,27 @@ Prometheus.Graph.prototype.updateGraph = function() {
     min: "auto",
   });
 
+  // Find and set graph's max/min
+  var min = Infinity;
+  var max = -Infinity;
+  self.data.forEach(function(timeSeries) {
+    timeSeries.data.forEach(function(dataPoint) {
+        if (dataPoint.y < min && dataPoint.y != null) {
+          min = dataPoint.y;
+        }
+        if (dataPoint.y > max && dataPoint.y != null) {
+          max = dataPoint.y;
+        }
+    });
+  });
+  if (min === max) {
+    self.rickshawGraph.max = max + 1;
+    self.rickshawGraph.min = min - 1;
+  } else {
+    self.rickshawGraph.max = max + (0.1*(Math.abs(max - min)));
+    self.rickshawGraph.min = min - (0.1*(Math.abs(max - min)));
+  }
+
   var xAxis = new Rickshaw.Graph.Axis.Time({ graph: self.rickshawGraph });
 
   var yAxis = new Rickshaw.Graph.Axis.Y({

web/ui/static/js/prom_console.js

@@ -578,7 +578,7 @@ PromConsole.Graph.prototype.dispatch = function() {
   }
 
   var loadingImg = document.createElement("img");
-  loadingImg.src = PATH_PREFIX + '/static/img/ajax-loader.gif?v=' + BUILD_VERSION;
+  loadingImg.src = PATH_PREFIX + '/static/img/ajax-loader.gif';
   loadingImg.alt = 'Loading...';
   loadingImg.className = 'prom_graph_loading';
   this.graphTd.appendChild(loadingImg);

web/ui/templates/status.html

@@ -54,7 +54,8 @@
         </tr>
         {{range .Alertmanagers}}
         <tr>
-          <td>{{.}}</td>
+          {{/* Alertmanager URLs always have Scheme, Host and Path set */}}
+          <td>{{.Scheme}}://<a href="{{.Scheme}}://{{.Host}}">{{.Host}}</a>{{.Path}}</td>
         </tr>
         {{end}}
       </tbody>

web/web.go

@@ -348,7 +348,7 @@ func (h *Handler) status(w http.ResponseWriter, r *http.Request) {
 		Birth         time.Time
 		CWD           string
 		Version       *PrometheusVersion
-		Alertmanagers []string
+		Alertmanagers []*url.URL
 	}{
 		Birth:         h.birth,
 		CWD:           h.cwd,