Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

*: add pending task metrics in runner and add templ #8138

Merged
merged 7 commits into from
May 10, 2024
Merged

*: add pending task metrics in runner and add templ #8138

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
23fdf1c
*: add pending task metrics in runner and add templ
nolouch Apr 28, 2024
2e89823
address
nolouch Apr 30, 2024
d1590fb
Merge branch 'master' into add-metrics
nolouch Apr 30, 2024
4228985
Merge branch 'master' into add-metrics
nolouch May 6, 2024
feaac64
Merge remote-tracking branch 'origin/master' into add-metrics
nolouch May 9, 2024
c068dca
Merge remote-tracking branch 'nolouch/add-metrics' into add-metrics
nolouch May 9, 2024
b35c0ce
address
nolouch May 10, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
256 changes: 248 additions & 8 deletions metrics/grafana/pd.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11606,17 +11606,257 @@
"dashLength": 10,
"dashes": false,
"datasource": "${DS_TEST-CLUSTER}",
"description": "The count of the corresponding schedule commands which PD sends to each TiKV instance",
"description": "The count of the heartbeats which pending in the task queue.",
"editable": true,
"error": false,
"fieldConfig": {
"defaults": {},
"overrides": []
},
"fill": 0,
"fillGradient": 0,
"grid": {},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 39
},
"hiddenSeries": false,
"id": 1608,
"legend": {
"alignAsTable": true,
"avg": true,
"current": true,
"hideEmpty": true,
"hideZero": true,
"max": true,
"min": false,
"rightSide": true,
"show": true,
"total": false,
"values": true
},
"lines": true,
"linewidth": 1,
"links": [],
"nullPointMode": "null as zero",
"options": {
"alertThreshold": true
},
"paceLength": 10,
"percentage": false,
"pluginVersion": "7.5.17",
"pointradius": 5,
"points": false,
"renderer": "flot",
"seriesOverrides": [],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"exemplar": true,
"expr": "pd_ratelimit_runner_task_pending_tasks{k8s_cluster=\"$k8s_cluster\", tidb_cluster=\"$tidb_cluster\"}",
"format": "time_series",
"hide": false,
"interval": "",
"intervalFactor": 2,
"legendFormat": "{{task_type}}_({{ runner_name}})",
nolouch marked this conversation as resolved.
Show resolved Hide resolved
"refId": "A",
"step": 4
}
],
"thresholds": [],
"timeFrom": null,
"timeRegions": [],
"timeShift": null,
"title": "Heartbeat Runner Pending Task",
"tooltip": {
"msResolution": false,
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": []
},
"yaxes": [
{
"$$hashKey": "object:505",
nolouch marked this conversation as resolved.
Show resolved Hide resolved
"format": "opm",
"label": null,
"logBase": 1,
"max": null,
"min": "0",
"show": true
},
{
"$$hashKey": "object:506",
"format": "s",
"label": null,
"logBase": 1,
"max": null,
"min": null,
"show": true
}
],
"yaxis": {
"align": false,
"alignLevel": null
}
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "${DS_TEST-CLUSTER}",
"description": "The count of the heartbeats which faileds in the task queue.",
"editable": true,
"error": false,
"fieldConfig": {
"defaults": {},
"overrides": []
},
"fill": 0,
"fillGradient": 0,
"grid": {},
"gridPos": {
"h": 8,
"w": 12,
"x": 0,
"y": 47
},
"hiddenSeries": false,
"id": 1609,
"legend": {
"alignAsTable": true,
"avg": true,
"current": true,
"hideEmpty": true,
"hideZero": true,
"max": true,
"min": false,
"rightSide": true,
"show": true,
"total": false,
"values": true
},
"lines": true,
"linewidth": 1,
"links": [],
"nullPointMode": "null as zero",
"options": {
"alertThreshold": true
},
"paceLength": 10,
"percentage": false,
"pluginVersion": "7.5.17",
"pointradius": 5,
"points": false,
"renderer": "flot",
"seriesOverrides": [
{
"$$hashKey": "object:243",
"alias": "/max-wait-duration.*/",
"bars": true,
"lines": false,
"transform": "negative-Y",
"yaxis": 2
}
],
"spaceLength": 10,
"stack": false,
"steppedLine": false,
"targets": [
{
"exemplar": true,
"expr": "rate(pd_ratelimit_runner_task_failed_tasks_total{k8s_cluster=\"$k8s_cluster\", tidb_cluster=\"$tidb_cluster\"}[1m])*60",
"format": "time_series",
"hide": false,
"interval": "",
"intervalFactor": 2,
"legendFormat": "failed-tasks-({{runner_name}})",
"refId": "A",
"step": 4
},
{
"exemplar": true,
"expr": "pd_ratelimit_runner_task_max_waiting_duration_seconds{k8s_cluster=\"$k8s_cluster\", tidb_cluster=\"$tidb_cluster\"}",
"hide": false,
"interval": "",
"legendFormat": "max-wait-duration-({{runner_name}})",
"refId": "B"
}
],
"thresholds": [],
"timeFrom": null,
"timeRegions": [],
"timeShift": null,
"title": "Concurrent Runner Failed Task",
"tooltip": {
"msResolution": false,
"shared": true,
"sort": 0,
"value_type": "individual"
},
"type": "graph",
"xaxis": {
"buckets": null,
"mode": "time",
"name": null,
"show": true,
"values": []
},
"yaxes": [
{
"$$hashKey": "object:201",
"decimals": null,
"format": "opm",
"label": "",
"logBase": 1,
"max": null,
"min": "0",
"show": true
},
{
"$$hashKey": "object:202",
"format": "s",
"label": null,
"logBase": 1,
"max": null,
"min": null,
"show": true
}
],
"yaxis": {
"align": false,
"alignLevel": null
}
},
{
"aliasColors": {},
"bars": false,
"dashLength": 10,
"dashes": false,
"datasource": "${DS_TEST-CLUSTER}",
"description": "The count of the corresponding schedule commands which PD sends to each TiKV instance",
"editable": true,
"error": false,
"fill": 0,
"grid": {},
"gridPos": {
"h": 8,
"w": 12,
"x": 12,
"y": 47
},
"id": 1305,
"legend": {
"alignAsTable": true,
Expand Down Expand Up @@ -11709,7 +11949,7 @@
"h": 8,
"w": 12,
"x": 0,
"y": 47
"y": 55
},
"id": 1306,
"legend": {
Expand Down Expand Up @@ -11799,7 +12039,7 @@
"h": 8,
"w": 12,
"x": 12,
"y": 47
"y": 55
},
"id": 1307,
"legend": {
Expand Down Expand Up @@ -11892,7 +12132,7 @@
"h": 8,
"w": 12,
"x": 0,
"y": 55
"y": 63
},
"id": 1308,
"legend": {
Expand Down Expand Up @@ -11989,7 +12229,7 @@
"h": 8,
"w": 12,
"x": 12,
"y": 55
"y": 63
},
"id": 1309,
"legend": {
Expand Down Expand Up @@ -12086,7 +12326,7 @@
"h": 8,
"w": 12,
"x": 0,
"y": 63
"y": 71
},
"id": 1310,
"legend": {
Expand Down Expand Up @@ -12183,7 +12423,7 @@
"h": 8,
"w": 12,
"x": 12,
"y": 63
"y": 71
},
"id": 1311,
"legend": {
Expand Down Expand Up @@ -12280,7 +12520,7 @@
"h": 8,
"w": 12,
"x": 0,
"y": 71
"y": 79
},
"id": 1312,
"legend": {
Expand Down
7 changes: 5 additions & 2 deletions pkg/ratelimit/metrics.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,10 @@ import (
"github.com/prometheus/client_golang/prometheus"
)

const nameStr = "runner_name"
const (
nameStr = "runner_name"
taskStr = "task_type"
)

var (
RunnerTaskMaxWaitingDuration = prometheus.NewGaugeVec(
Expand All @@ -35,7 +38,7 @@ var (
Subsystem: "ratelimit",
Name: "runner_task_pending_tasks",
Help: "The number of pending tasks in the runner.",
}, []string{nameStr})
}, []string{nameStr, taskStr})
RunnerTaskFailedTasks = prometheus.NewCounterVec(
prometheus.CounterOpts{
Namespace: "pd",
Expand Down
7 changes: 7 additions & 0 deletions pkg/ratelimit/runner.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,6 +54,7 @@ type ConcurrentRunner struct {
pendingMu sync.Mutex
stopChan chan struct{}
wg sync.WaitGroup
pendingTaskCount map[string]int64
failedTaskCount prometheus.Counter
maxWaitingDuration prometheus.Gauge
}
Expand All @@ -66,6 +67,7 @@ func NewConcurrentRunner(name string, maxPendingDuration time.Duration) *Concurr
taskChan: make(chan *Task),
pendingTasks: make([]*Task, 0, initialCapacity),
failedTaskCount: RunnerTaskFailedTasks.WithLabelValues(name),
pendingTaskCount: make(map[string]int64),
maxWaitingDuration: RunnerTaskMaxWaitingDuration.WithLabelValues(name),
}
return s
Expand Down Expand Up @@ -109,6 +111,9 @@ func (s *ConcurrentRunner) Start() {
if len(s.pendingTasks) > 0 {
maxDuration = time.Since(s.pendingTasks[0].submittedAt)
}
for name, cnt := range s.pendingTaskCount {
nolouch marked this conversation as resolved.
Show resolved Hide resolved
RunnerTaskPendingTasks.WithLabelValues(s.name, name).Set(float64(cnt))
}
s.pendingMu.Unlock()
s.maxWaitingDuration.Set(maxDuration.Seconds())
}
Expand All @@ -132,6 +137,7 @@ func (s *ConcurrentRunner) processPendingTasks() {
select {
case s.taskChan <- task:
s.pendingTasks = s.pendingTasks[1:]
s.pendingTaskCount[task.Opts.TaskName]--
return
default:
return
Expand Down Expand Up @@ -167,6 +173,7 @@ func (s *ConcurrentRunner) RunTask(ctx context.Context, opt TaskOpts, f func(con
}
task.submittedAt = time.Now()
s.pendingTasks = append(s.pendingTasks, task)
s.pendingTaskCount[opt.TaskName]++
}
return nil
}
Expand Down
2 changes: 1 addition & 1 deletion server/cluster/cluster.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -107,7 +107,7 @@ const (
minSnapshotDurationSec = 5

// heartbeat relative const
hbConcurrentRunner = "heartbeat-async-task-runner"
hbConcurrentRunner = "heartbeat-async"
)

// Server is the interface for cluster.
Expand Down