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

Support tidb-binlog metrics scraping for TidbMonitor #2750

Merged
merged 10 commits into from
Jun 19, 2020
Merged

Support tidb-binlog metrics scraping for TidbMonitor #2750

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
f523f0d
add pump
Jun 17, 2020
b1d9d56
add drainer
Jun 17, 2020
7f20a60
remove uselesscode
Jun 17, 2020
d02a270
fix unit test
Jun 17, 2020
3f3c16a
add tls
Jun 18, 2020
6970b51
Merge branch 'master' into scape_metrics_for_binlog
DanielZhangQD Jun 18, 2020
fb93c13
fix tls template
Jun 19, 2020
1e55a5a
fix tiflash
Jun 19, 2020
c063119
Merge branch 'master' into scape_metrics_for_binlog
ti-srebot Jun 19, 2020
71af806
fix monitor e2e bug
Jun 19, 2020
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
117 changes: 87 additions & 30 deletions pkg/monitor/monitor/template.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,17 +15,20 @@ package monitor

import (
"fmt"
"path"
"strings"
"time"

"github.com/pingcap/tidb-operator/pkg/util"
"github.com/prometheus/common/model"
"github.com/prometheus/prometheus/config"
"gopkg.in/yaml.v2"
corev1 "k8s.io/api/core/v1"
"k8s.io/klog"
"path"
"time"
)

const (
nameLabel = "__meta_kubernetes_pod_label_app_kubernetes_io_name"
instanceLabel = "__meta_kubernetes_pod_label_app_kubernetes_io_instance"
componentLabel = "__meta_kubernetes_pod_label_app_kubernetes_io_component"
scrapeLabel = "__meta_kubernetes_pod_annotation_prometheus_io_scrape"
Expand All @@ -45,6 +48,8 @@ var (
tidbPattern config.Regexp
addressPattern config.Regexp
tiflashPattern config.Regexp
pumpPattern config.Regexp
drainerPattern config.Regexp
dashBoardConfig = `{
"apiVersion": 1,
"providers": [
Expand Down Expand Up @@ -95,6 +100,14 @@ func init() {
if err != nil {
klog.Fatalf("monitor regex template parse error,%v", err)
}
pumpPattern, err = config.NewRegexp("pump")
if err != nil {
klog.Fatalf("monitor regex template parse error,%v", err)
}
drainerPattern, err = config.NewRegexp("drainer")
if err != nil {
klog.Fatalf("monitor regex template parse error,%v", err)
}
}

type MonitorConfigModel struct {
Expand All @@ -105,13 +118,6 @@ type MonitorConfigModel struct {
}

func newPrometheusConfig(cmodel *MonitorConfigModel) *config.Config {
pdReplacement := fmt.Sprintf("$1.$2-%s-peer:$3", "pd")
tikvReplacement := fmt.Sprintf("$1.$2-%s-peer:$3", "tikv")
tidbReplacement := fmt.Sprintf("$1.$2-%s-peer:$3", "tidb")
tiflashReplacement := fmt.Sprintf("$1.$2-%s-peer:$3", "tiflash")
tiflashProxyReplacement := fmt.Sprintf("$1.$2-%s-peer:$3", "tiflash")
tiflashProxyPortLabel := fmt.Sprintf(additionalPortLabelPattern, "tiflash_proxy")

var c = config.Config{
GlobalConfig: config.GlobalConfig{
ScrapeInterval: model.Duration(15 * time.Second),
Expand All @@ -121,41 +127,91 @@ func newPrometheusConfig(cmodel *MonitorConfigModel) *config.Config {
"/prometheus-rules/rules/*.rules.yml",
},
ScrapeConfigs: []*config.ScrapeConfig{
scrapeJob("pd", pdPattern, cmodel, buildAddressRelabelConfig(portLabel, pdReplacement, true)),
scrapeJob("tidb", tidbPattern, cmodel, buildAddressRelabelConfig(portLabel, tidbReplacement, true)),
scrapeJob("tikv", tikvPattern, cmodel, buildAddressRelabelConfig(portLabel, tikvReplacement, true)),
scrapeJob("tiflash", tiflashPattern, cmodel, buildAddressRelabelConfig(portLabel, tiflashReplacement, true)),
scrapeJob("tiflash-proxy", tiflashPattern, cmodel, buildAddressRelabelConfig(tiflashProxyPortLabel, tiflashProxyReplacement, true)),
scrapeJob("pd", pdPattern, cmodel, buildAddressRelabelConfigByComponent("pd")),
scrapeJob("tidb", tidbPattern, cmodel, buildAddressRelabelConfigByComponent("tidb")),
scrapeJob("tikv", tikvPattern, cmodel, buildAddressRelabelConfigByComponent("tikv")),
scrapeJob("tiflash", tiflashPattern, cmodel, buildAddressRelabelConfigByComponent("tiflash")),
scrapeJob("tiflash-proxy", tiflashPattern, cmodel, buildAddressRelabelConfigByComponent("tiflash-proxy")),
scrapeJob("pump", pumpPattern, cmodel, buildAddressRelabelConfigByComponent("pump")),
scrapeJob("drainer", drainerPattern, cmodel, buildAddressRelabelConfigByComponent("drainer")),
},
}
return &c
}

func buildAddressRelabelConfig(portLabelName, replacement string, isTidbClusterComponent bool) *config.RelabelConfig {
addressRelabelConfig := &config.RelabelConfig{
SourceLabels: model.LabelNames{
"__address__",
model.LabelName(portLabelName),
},
Action: config.RelabelReplace,
Regex: portPattern,
Replacement: "$1:$2",
TargetLabel: "__address__",
}
if isTidbClusterComponent {
addressRelabelConfig = &config.RelabelConfig{
func buildAddressRelabelConfigByComponent(kind string) *config.RelabelConfig {
kind = strings.ToLower(kind)
replacement := fmt.Sprintf("$1.$2-%s-peer:$3", kind)
f := func() *config.RelabelConfig {
return &config.RelabelConfig{
Action: config.RelabelReplace,
Regex: addressPattern,
Replacement: replacement,
TargetLabel: "__address__",
SourceLabels: model.LabelNames{
podNameLabel,
instanceLabel,
model.LabelName(portLabelName),
portLabel,
},
}
}
return addressRelabelConfig
switch strings.ToLower(kind) {
case "pd":
return f()
case "tidb":
return f()
case "tikv":
return f()
case "tiflash":
return f()
case "tiflash-proxy":
return &config.RelabelConfig{
Action: config.RelabelReplace,
Regex: addressPattern,
Replacement: "$1.$2-tiflash-peer:$3",
TargetLabel: "__address__",
SourceLabels: model.LabelNames{
podNameLabel,
instanceLabel,
model.LabelName(fmt.Sprintf(additionalPortLabelPattern, "tiflash_proxy")),
},
}
case "pump":
return &config.RelabelConfig{
Action: config.RelabelReplace,
Regex: addressPattern,
Replacement: "$1.$2-pump:$3",
TargetLabel: "__address__",
SourceLabels: model.LabelNames{
podNameLabel,
instanceLabel,
portLabel,
},
}
case "drainer":
return &config.RelabelConfig{
Action: config.RelabelReplace,
Regex: addressPattern,
Replacement: "$1.$2:$3",
TargetLabel: "__address__",
SourceLabels: model.LabelNames{
podNameLabel,
nameLabel,
portLabel,
},
}
default:
return &config.RelabelConfig{
SourceLabels: model.LabelNames{
"__address__",
portLabel,
},
Action: config.RelabelReplace,
Regex: portPattern,
Replacement: "$1:$2",
TargetLabel: "__address__",
}
}
}

func scrapeJob(jobName string, componentPattern config.Regexp, cmodel *MonitorConfigModel, addressRelabelConfig *config.RelabelConfig) *config.ScrapeConfig {
Expand Down Expand Up @@ -273,7 +329,8 @@ func addTlsConfig(pc *config.Config) {

for id, sconfig := range pc.ScrapeConfigs {
// TODO support tiflash tls when it gets ready
if sconfig.JobName == "pd" || sconfig.JobName == "tidb" || sconfig.JobName == "tikv" {
if sconfig.JobName == "pd" || sconfig.JobName == "tidb" || sconfig.JobName == "tikv" ||
sconfig.JobName == "pump" || sconfig.JobName == "drainer" {
sconfig.HTTPClientConfig.TLSConfig = config.TLSConfig{
CAFile: path.Join(util.ClusterClientTLSPath, corev1.ServiceAccountRootCAKey),
CertFile: path.Join(util.ClusterClientTLSPath, corev1.TLSCertKey),
Expand Down
198 changes: 198 additions & 0 deletions pkg/monitor/monitor/template_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -273,6 +273,102 @@ scrape_configs:
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
- job_name: pump
honor_labels: true
scrape_interval: 15s
scheme: http
kubernetes_sd_configs:
- api_server: null
role: pod
namespaces:
names:
- ns1
- ns2
tls_config:
insecure_skip_verify: true
relabel_configs:
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
regex: target
action: keep
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_component]
regex: pump
action: keep
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_scrape]
regex: "true"
action: keep
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_path]
regex: (.+)
target_label: __metrics_path__
action: replace
- source_labels: [__meta_kubernetes_pod_name, __meta_kubernetes_pod_label_app_kubernetes_io_instance,
__meta_kubernetes_pod_annotation_prometheus_io_port]
regex: (.+);(.+);(.+)
target_label: __address__
replacement: $1.$2-pump:$3
action: replace
- source_labels: [__meta_kubernetes_namespace]
target_label: kubernetes_namespace
action: replace
- source_labels: [__meta_kubernetes_pod_name]
target_label: instance
action: replace
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
- source_labels: [__meta_kubernetes_pod_name]
target_label: instance
action: replace
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
- job_name: drainer
honor_labels: true
scrape_interval: 15s
scheme: http
kubernetes_sd_configs:
- api_server: null
role: pod
namespaces:
names:
- ns1
- ns2
tls_config:
insecure_skip_verify: true
relabel_configs:
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
regex: target
action: keep
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_component]
regex: drainer
action: keep
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_scrape]
regex: "true"
action: keep
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_path]
regex: (.+)
target_label: __metrics_path__
action: replace
- source_labels: [__meta_kubernetes_pod_name, __meta_kubernetes_pod_label_app_kubernetes_io_name,
__meta_kubernetes_pod_annotation_prometheus_io_port]
regex: (.+);(.+);(.+)
target_label: __address__
replacement: $1.$2:$3
action: replace
- source_labels: [__meta_kubernetes_namespace]
target_label: kubernetes_namespace
action: replace
- source_labels: [__meta_kubernetes_pod_name]
target_label: instance
action: replace
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
- source_labels: [__meta_kubernetes_pod_name]
target_label: instance
action: replace
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
`
model := &MonitorConfigModel{
ReleaseTargetRegex: &target,
Expand Down Expand Up @@ -546,6 +642,108 @@ scrape_configs:
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
- job_name: pump
honor_labels: true
scrape_interval: 15s
scheme: https
kubernetes_sd_configs:
- api_server: null
role: pod
namespaces:
names:
- ns1
- ns2
tls_config:
ca_file: /var/lib/cluster-client-tls/ca.crt
cert_file: /var/lib/cluster-client-tls/tls.crt
key_file: /var/lib/cluster-client-tls/tls.key
insecure_skip_verify: false
relabel_configs:
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
regex: target
action: keep
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_component]
regex: pump
action: keep
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_scrape]
regex: "true"
action: keep
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_path]
regex: (.+)
target_label: __metrics_path__
action: replace
- source_labels: [__meta_kubernetes_pod_name, __meta_kubernetes_pod_label_app_kubernetes_io_instance,
__meta_kubernetes_pod_annotation_prometheus_io_port]
regex: (.+);(.+);(.+)
target_label: __address__
replacement: $1.$2-pump:$3
action: replace
- source_labels: [__meta_kubernetes_namespace]
target_label: kubernetes_namespace
action: replace
- source_labels: [__meta_kubernetes_pod_name]
target_label: instance
action: replace
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
- source_labels: [__meta_kubernetes_pod_name]
target_label: instance
action: replace
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
- job_name: drainer
honor_labels: true
scrape_interval: 15s
scheme: https
kubernetes_sd_configs:
- api_server: null
role: pod
namespaces:
names:
- ns1
- ns2
tls_config:
ca_file: /var/lib/cluster-client-tls/ca.crt
cert_file: /var/lib/cluster-client-tls/tls.crt
key_file: /var/lib/cluster-client-tls/tls.key
insecure_skip_verify: false
relabel_configs:
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
regex: target
action: keep
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_component]
regex: drainer
action: keep
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_scrape]
regex: "true"
action: keep
- source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_path]
regex: (.+)
target_label: __metrics_path__
action: replace
- source_labels: [__meta_kubernetes_pod_name, __meta_kubernetes_pod_label_app_kubernetes_io_name,
__meta_kubernetes_pod_annotation_prometheus_io_port]
regex: (.+);(.+);(.+)
target_label: __address__
replacement: $1.$2:$3
action: replace
- source_labels: [__meta_kubernetes_namespace]
target_label: kubernetes_namespace
action: replace
- source_labels: [__meta_kubernetes_pod_name]
target_label: instance
action: replace
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
- source_labels: [__meta_kubernetes_pod_name]
target_label: instance
action: replace
- source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
target_label: cluster
action: replace
`
model := &MonitorConfigModel{
ReleaseTargetRegex: &target,
Expand Down
Loading