使用 Prometheus 监控 Harbor
你好！我是李大白，今天分享的是基于Prometheus监控harbor服务。
在之前的文章中分别介绍了harbor基于离线安装的高可用汲取设计和部署。那么，如果我们的harbor服务主机或者harbor服务及组件出现异常，我们该如何快速处理呢？
Harbor v2.2及以上版本支持配置Prometheus监控Harbor，所以你的harbor版本必须要大于2.2。
本篇文章以二进制的方式简单的部署Prometheus相关服务，可以帮助你快速的的实现Prometheus对harbor的监控。
Prometheus监控Harbor（二进制版）
一、部署说明
在harbor服务主机上部署:
prometheus node-exporter grafana alertmanager
harbor版本：2.4.2
主机：192.168.2.22
二、Harbor启用metrics服务
2.1 停止Harbor服务
$ cd /app/harbor
$ docker-compose down
2.2 修改harbor.yml配置
修改harbor的配置文件中metrics参数，启用harbor-exporter组件。
$ cat harbor.yml
### metrics配置部分
metric:
enabled: true #是否启用，需要修改为true（启用）
port: 9099 #默认的端口为9090，与prometheus的端口会冲突（所以需要修改下）
path: /metrics
对harbor不熟悉的建议对配置文件备份下！
2.3 配置注入组件
$ ./prepre
2.4 install安装harbor
$ ./install.sh --with-notary --with-trivy --with-chartmuseum
$ docker-compose ps
NAME COMMAND SERVICE STATUS PORTS
chartmuseum "./docker-entrypoint…" chartmuseum running (healthy)
harbor-core "/harbor/entrypoint.…" core running (healthy)
harbor-db "/docker-entrypoint.…" postgresql running (healthy)
harbor-exporter "/harbor/entrypoint.…" exporter running
可以看到多了harbor-exporter组件。
三、Harbor指标说明
在前面启用了harbor-exporter监控组件后，可以通过curl命令去查看harbor暴露了哪些指标。
harbor暴露了以下4个关键组件的指标数据。
3.1 harbor-exporter组件指标
exporter组件指标与Harbor 实例配置相关，并从 Harbor 数据库中收集一些数据。指标可在
<harbor_instance>:<metrics_port>/<metrics_path>查看
$ curl http://192.168.2.22:9099/metrics
1）harbor_project_total
harbor_project_total 采集了公共和私人项目总共数量。
$ curl http://192.168.2.22:9099/metrics | grep harbor_project_total
# HELP harbor_project_total Total projects number
# TYPE harbor_project_total gauge
harbor_project_total{public="true"} 1 # 公共项目的数量为“1”
harbor_project_total{public="false"} 1 #私有项目的数量
2）harbor_project_repo_total
项目（Project）中的存储库总数。
$ curl http://192.168.2.22:9099/metrics | grep harbor_project_repo_total
# HELP harbor_project_repo_total Total project repos number
# TYPE harbor_project_repo_total gauge
harbor_project_repo_total{project_name="library",public="true"} 0
3）harbor_project_member_total
项目中的成员总数
$ curl http://192.168.2.22:9099/metrics | grep harbor_project_member_total
# HELP harbor_project_member_total Total members number of a project
# TYPE harbor_project_member_total gauge
harbor_project_member_total{project_name="library"} 1 #项目library下有“1”个用户
4）harbor_project_quota_usage_byte
一个项目的总使用资源
$ curl http://192.168.2.22:9099/metrics | grep harbor_project_quota_usage_byte
# HELP harbor_project_quota_usage_byte The used resource of a project
# TYPE harbor_project_quota_usage_byte gauge
harbor_project_quota_usage_byte{project_name="library"} 0
5）harbor_project_quota_byte
项目中设置的配额
$ curl http://192.168.2.22:9099/metrics | grep harbor_project_quota_byte
# HELP harbor_project_quota_byte The quota of a project
# TYPE harbor_project_quota_byte gauge
harbor_project_quota_byte{project_name="library"} -1 #-1 表示不限制
6）harbor_artifact_pulled
项目中镜像拉取的总数
$ curl http://192.168.2.22:9099/metrics | grep harbor_artifact_pulled
# HELP harbor_artifact_pulled The pull number of an artifact
# TYPE harbor_artifact_pulled gauge
harbor_artifact_pulled{project_name="library"} 0
7）harbor_project_artifact_total
项目中的工件类型总数，artifact_type , project_name, public ( true, false)
$ curl http://192.168.2.22:9099/metrics | grep harbor_project_artifact_total
8）harbor_health
Harbor状态$ curl http://192.168.2.22:9099/metrics | grep harbor_health
# HELP harbor_health Running status of Harbor
# TYPE harbor_health gauge
harbor_health 1 #1表示正常，0表示异常
9）harbor_system_info
Harbor 实例的信息，auth_mode ( db_auth, ldap_auth, uaa_auth, http_auth, oidc_auth),harbor_version, self_registration( true, false)
$ curl http://192.168.2.22:9099/metrics | grep harbor_system_info
# HELP harbor_system_info Information of Harbor system
# TYPE harbor_system_info gauge
harbor_system_info{auth_mode="db_auth",harbor_version="v2.4.2-ef2e2e56",self_registration="false"} 1
10）harbor_up
Harbor组件运行状态，组件 ( chartmuseum, core, database, jobservice, portal, redis, registry, registryctl, trivy)
$ curl http://192.168.2.22:9099/metrics | grep harbor_up
harbor_up Running status of harbor component
# TYPE harbor_up gauge
harbor_up{component="chartmuseum"} 1
harbor_up{component="core"} 1
harbor_up{component="database"} 1
harbor_up{component="jobservice"} 1
harbor_up{component="portal"} 1
harbor_up{component="redis"} 1
harbor_up{component="registry"} 1
harbor_up{component="registryctl"} 1
harbor_up{component="trivy"} 1 #Trivy扫描器运行状态
11）harbor_task_queue_size
队列中每种类型的任务总数，
$ curl http://192.168.2.22:9099/metrics | grep harbor_task_queue_size
# HELP harbor_task_queue_size Total number of tasks
# TYPE harbor_task_queue_size gauge
harbor_task_queue_size{type="DEMO"} 0
harbor_task_queue_size{type="GARBAGE_COLLECTION"} 0
harbor_task_queue_size{type="IMAGE_GC"} 0
harbor_task_queue_size{type="IMAGE_REPLICATE"} 0
harbor_task_queue_size{type="IMAGE_SCAN"} 0
harbor_task_queue_size{type="IMAGE_SCAN_ALL"} 0
harbor_task_queue_size{type="P2P_PREHEAT"} 0
harbor_task_queue_size{type="REPLICATION"} 0
harbor_task_queue_size{type="RETENTION"} 0
harbor_task_queue_size{type="SCHEDULER"} 0
harbor_task_queue_size{type="SLACK"} 0
harbor_task_queue_size{type="WEBHOOK"} 0
12）harbor_task_queue_latency
多久前要处理的下一个作业按类型排入队列
$ curl http://192.168.2.22:9099/metrics | grep harbor_task_queue_latency
# HELP harbor_task_queue_latency how long ago the next job to be processed was enqueued
# TYPE harbor_task_queue_latency gauge
harbor_task_queue_latency{type="DEMO"} 0
harbor_task_queue_latency{type="GARBAGE_COLLECTION"} 0
harbor_task_queue_latency{type="IMAGE_GC"} 0
harbor_task_queue_latency{type="IMAGE_REPLICATE"} 0
harbor_task_queue_latency{type="IMAGE_SCAN"} 0
harbor_task_queue_latency{type="IMAGE_SCAN_ALL"} 0
harbor_task_queue_latency{type="P2P_PREHEAT"} 0
harbor_task_queue_latency{type="REPLICATION"} 0
harbor_task_queue_latency{type="RETENTION"} 0
harbor_task_queue_latency{type="SCHEDULER"} 0
harbor_task_queue_latency{type="SLACK"} 0
harbor_task_queue_latency{type="WEBHOOK"} 0
13）harbor_task_scheduled_total
计划任务数
$ curl http://192.168.2.22:9099/metrics | grep harbor_task_scheduled_total
# HELP harbor_task_scheduled_total total number of scheduled job
# TYPE harbor_task_scheduled_total gauge
harbor_task_scheduled_total 0
14）harbor_task_concurrency
池（Total)上每种类型的并发任务总数
$ curl http://192.168.2.22:9099/metrics | grep harbor_task_concurrency
harbor_task_concurrency{pool="d4053262b74f0a7b83bc6add",type="GARBAGE_COLLECTION"} 0
3.2 harbor-core组件指标
以下是从 Harbor core组件中提取的指标，获取格式：
<harbor_instance>:<metrics_port>/<metrics_path>?comp=core.
1）harbor_core_http_inflight_requests
请求总数，操作（Harbor API operationId中的值。一些遗留端点没有，因此标签值为）operationId``unknown
harbor-core组件的指标
$ curl http://192.168.2.22:9099/metrics?comp=core | grep harbor_core_http_inflight_requests
# HELP harbor_core_http_inflight_requests The total number of requests
# TYPE harbor_core_http_inflight_requests gauge
harbor_core_http_inflight_requests 0
2）harbor_core_http_request_duration_seconds
请求的持续时间，
方法 ( GET, POST, HEAD, PATCH, PUT), 操作 ( Harbor APIoperationId中的 值。一些遗留端点没有, 所以标签值为), 分位数operationId``unknown
$ curl http://192.168.2.22:9099/metrics?comp=core | grep harbor_core_http_request_duration_seconds
# HELP harbor_core_http_request_duration_seconds The time duration of the requests
# TYPE harbor_core_http_request_duration_seconds summary
harbor_core_http_request_duration_seconds{method="GET",operation="GetHealth",quantile="0.5"} 0.001797115
harbor_core_http_request_duration_seconds{method="GET",operation="GetHealth",quantile="0.9"} 0.010445204
harbor_core_http_request_duration_seconds{method="GET",operation="GetHealth",quantile="0.99"} 0.010445204
3）harbor_core_http_request_total
请求总数
方法（GET, POST, HEAD, PATCH, PUT），操作（[Harbor API operationId中的 值。一些遗留端点没有，因此标签值为）operationId``unknown
$ curl http://192.168.2.22:9099/metrics?comp=core | grep harbor_core_http_request_total
# HELP harbor_core_http_request_total The total number of requests
# TYPE harbor_core_http_request_total counter
harbor_core_http_request_total{code="200",method="GET",operation="GetHealth"} 14
harbor_core_http_request_total{code="200",method="GET",operation="GetInternalconfig"} 1
harbor_core_http_request_total{code="200",method="GET",operation="GetPing"} 176
harbor_core_http_request_total{code="200",method="GET",operation="GetSystemInfo"} 14
3.3 registry 组件指标
注册表，以下是从 Docker 发行版中提取的指标，查看指标方式：
<harbor_instance>:<metrics_port>/<metrics_path>?comp=registry.
1）registry_http_in_flight_requests
进行中的 HTTP 请求，处理程序
$ curl http://192.168.2.22:9099/metrics?comp=registry | grep registry_http_in_flight_requests
# HELP registry_http_in_flight_requests The in-flight HTTP requests
# TYPE registry_http_in_flight_requests gauge
registry_http_in_flight_requests{handler="base"} 0
registry_http_in_flight_requests{handler="blob"} 0
registry_http_in_flight_requests{handler="blob_upload"} 0
registry_http_in_flight_requests{handler="blob_upload_chunk"} 0
registry_http_in_flight_requests{handler="catalog"} 0
registry_http_in_flight_requests{handler="manifest"} 0
registry_http_in_flight_requests{handler="tags"} 0
2）registry_http_request_duration_seconds
HTTP 请求延迟（以秒为单位），处理程序、方法( ,,,, GET) POST,文件HEADPATCHPUT
$ curl http://192.168.2.22:9099/metrics?comp=registry | grep registry_http_request_duration_seconds
3）registry_http_request_size_bytes
HTTP 请求大小（以字节为单位）。
$ curl http://192.168.2.22:9099/metrics?comp=registry | grep registry_http_request_size_bytes
3.4 jobservice组件指标
以下是从 Harbor Jobservice 提取的指标，
可在<harbor_instance>:<metrics_port>/<metrics_path>?comp=jobservice.查看
1）harbor_jobservice_info
Jobservice的信息,
$ curl http://192.168.2.22:9099/metrics?comp=jobservice | grep harbor_jobservice_info
# HELP harbor_jobservice_info the information of jobservice
# TYPE harbor_jobservice_info gauge
harbor_jobservice_info{node="f47de52e23b7:172.18.0.11",pool="35f1301b0e261d18fac7ba41",workers="10"} 1
2）harbor_jobservice_task_total
每个作业类型处理的任务数
$ curl http://192.168.2.22:9099/metrics?comp=jobservice | grep harbor_jobservice_task_tota
3）harbor_jobservice_task_process_time_seconds
任务处理时间的持续时间，即任务从开始执行到任务结束用了多少时间。
$ curl http://192.168.2.22:9099/metrics?comp=jobservice | grep harbor_jobservice_task_process_time_seconds
四、部署Prometheus Server（二进制）
4.1 创建安装目录
$ mkdir /etc/prometheus
4.2 下载安装包
$ wget https://github.com/prometheus/prometheus/releases/download/v2.36.2/prometheus-2.36.2.linux-amd64.tar.gz -c
$ tar zxvf prometheus-2.36.2.linux-amd64.tar.gz -C /etc/prometheus
$ cp prometheus-2.36.2.linux-amd64/{prometheus,promtool} /usr/local/bin/
$ prometheus --version #查看版本
prometheus, version 2.36.2 (branch: HEAD, revision: d7e7b8e04b5ecdc1dd153534ba376a622b72741b)
build user: root@f051ce0d6050
build date: 20220620-13:21:35
go version: go1.18.3
platform: linux/amd64
4.3 修改配置文件
在prometheus的配置文件中指定获取harbor采集的指标数据。
$ cp prometheus-2.36.2.linux-amd64/prometheus.yml /etc/prometheus/
$ cat <<EOF > /etc/prometheus/prometheus.yml
global:
scrape_interval: 15s
evaluation_interval: 15s
## 指定Alertmanagers地址
alerting:
alertmanagers:
- static_configs:
- targets: ["192.168.2.10:9093"] #填写Alertmanagers地址
## 配置告警规则文件
rule_files: #指定告警规则
- /etc/prometheus/rules.yml
scrape_configs:
- job_name: "prometheus"
static_configs:
- targets: ["localhost:9090"]
- job_name: 'node-exporter'
static_configs:
- targets:
- '192.168.2.22:9100'
- job_name: "harbor-exporter"
scrape_interval: 20s
static_configs:
- targets: ['192.168.2.22:9099']
- job_name: 'harbor-core'
params:
comp: ['core']
static_configs:
- targets: ['192.168.2.22:9099']
- job_name: 'harbor-registry'
params:
comp: ['registry']
static_configs:
- targets: ['192.168.2.22:9099']
- job_name: 'harbor-jobservice'
params:
comp: ['jobservice']
static_configs:
- targets: ['192.168.2.22:9099']
EOF
4.4 语法检查
检测配置文件的语法是否正确！
$ promtool check config /etc/prometheus/prometheus.yml
Checking /etc/prometheus/prometheus.yml
SUCCESS: /etc/prometheus/prometheus.yml is valid prometheus config file syntax
Checking /etc/prometheus/rules.yml
SUCCESS: 6 rules found
4.5 创建服务启动文件
$ cat <<EOF > /usr/lib/systemd/system/prometheus.service
[Unit]
Description=Prometheus Service
Documentation=https://prometheus.io/docs/introduction/overview/
wants=network-online.target
After=network-online.target
[Service]
Type=simple
User=root
Group=root
ExecStart=/usr/local/bin/prometheus --config.file=/etc/prometheus/prometheus.yml
[Install]
WantedBy=multi-user.target
EOF
4.6 启动服务
$ systemctl daemon-reload
$ systemctl enable --now prometheus.service
$ systemctl status prometheus.service
4.7 浏览器访问Prometheus UI
在浏览器地址栏输入主机IP:9090访问Prometheus UI 管理界面。
五、部署node-exporter
node-exporter服务可采集主机的cpu、内存、磁盘等资源指标。
5.1 下载安装包
$ wget https://github.com/prometheus/node_exporter/releases/download/v1.2.2/node_exporter-1.2.2.linux-amd64.tar.gz
$ tar zxvf node_exporter-1.2.2.linux-amd64.tar.gz
$ cp node_exporter-1.2.2.linux-amd64/node_exporter /usr/local/bin/
$ node_exporter --version
node_exporter, version 1.2.2 (branch: HEAD, revision: 26645363b486e12be40af7ce4fc91e731a33104e)
build user: root@b9cb4aa2eb17
build date: 20210806-13:44:18
go version: go1.16.7
platform: linux/amd64
5.2 创建服务启动文件
$ cat <<EOF > /usr/lib/systemd/system/node-exporter.service
[Unit]
Description=Prometheus Node Exporter
After=network.target
[Service]
ExecStart=/usr/local/bin/node_exporter
#User=prometheus
[Install]
WantedBy=multi-user.target
EOF
5.3 启动服务
$ systemctl daemon-reload
$ systemctl enable --now node-exporter.service
$ systemctl status node-exporter.service
$ ss -ntulp | grep node_exporter
tcp LISTEN 0 128 :::9100 :::* users:(("node_exporter",pid=36218,fd=3)
5.4 查看node指标
通过curl获取node-exporter服务采集到的监控数据。
$ curl http://localhost:9100/metrics
六、Grafana部署与仪表盘设计
二进制部署Grafana v8.4.4服务。
6.1 下载安装包
$ wget https://dl.grafana.com/enterprise/release/grafana-enterprise-8.4.4.linux-amd64.tar.gz -c
$ tar zxvf grafana-enterprise-8.4.4.linux-amd64.tar.gz -C /etc/
$ mv /etc/grafana-8.4.4 /etc/grafana
$ cp -a /etc/grafana/bin/{grafana-cli,grafana-server} /usr/local/bin/
#安装依赖包
$ yum install -y fontpackages-filesystem.noarch libXfont libfontenc lyx-fonts.noarch xorg-x11-font-utils
6.2 安装插件
安装grafana时钟插件
$ grafana-cli plugins install grafana-clock-panel
安装Zabbix插件
$ grafana-cli plugins install alexanderzobnin-zabbix-app
安装服务器端图像渲染组件
$ yum install -y fontconfig freetype* urw-fonts
6.3 创建服务启动文件
$ cat <<EOF >/usr/lib/systemd/system/grafana.service
[Service]
Type=notify
ExecStart=/usr/local/bin/grafana-server -homepath /etc/grafana
Restart=on-failure
[Install]
WantedBy=multi-user.target
EOF
-homepath：指定grafana的工作目录
6.4 启动grafana服务
$ systemctl daemon-reload
$ systemctl enable --now grafana.service
$ systemctl status grafana.service
$ ss -ntulp | grep grafana-server
tcp LISTEN 0 128 :::3000 :::* users:(("grafana-server",pid=120140,fd=9))
6.5 配置数据源
在浏览器地址栏输入主机IP和grafana服务端口访问Grafana UI界面后，添加Prometheus数据源。
默认用户密码：admin/admin
6.6 导入json模板
一旦您配置了Prometheus服务器以收集您的 Harbor 指标，您就可以使用 Grafana来可视化您的数据。Harbor 存储库中提供了一个 示例 Grafana 仪表板，可帮助您开始可视化 Harbor 指标。
Harbor官方提供了一个grafana的json文件模板。下载：
https://github.com/goharbor/harbor/blob/main/contrib/grafana-dashborad/metrics-example.json
七、部署AlertManager服务（扩展）
Alertmanager是一个独立的告警模块，接收Prometheus等客户端发来的警报，之后通过分组、删除重复等处理，并将它们通过路由发送给正确的接收器；
7.1 下载安装包
$ wget https://github.com/prometheus/alertmanager/releases/download/v0.23.0/alertmanager-0.23.0.linux-amd64.tar.gz
$ tar zxvf alertmanager-0.23.0.linux-amd64.tar.gz
$ cp alertmanager-0.23.0.linux-amd64/{alertmanager,amtool} /usr/local/bin/
7.2 修改配置文件
$ mkdir /etc/alertmanager
$ cat /etc/alertmanager/alertmanager.yml
global:
resolve_timeout: 5m
route:
group_by: ['alertname']
group_wait: 10s
group_interval: 10s
repeat_interval: 1h
receiver: 'web.hook'
receivers:
- name: 'web.hook'
webhook_configs:
- url: 'http://127.0.0.1:5001/'
inhibit_rules:
- source_match:
severity: 'critical'
target_match:
severity: 'warning'
equal: ['alertname', 'dev', 'instance']
7.3 创建服务启动文件
$ cat <<EOF >/usr/lib/systemd/system/alertmanager.service
[Unit]
Description=alertmanager
fter=network.target
[Service]
ExecStart=/usr/local/bin/alertmanager --config.file=/etc/alertmanager/alertmanager.yml
ExecReload=/bin/kill -HUP $MAINPID
KillMode=process
Restart=on-failure
[Install]
WantedBy=multi-user.target
EOF
7.4 启动服务
$ systemctl daemon-reload
$ systemctl enable --now alertmanager.service
$ systemctl status alertmanager.service
$ ss -ntulp | grep alertmanager
7.5 配置告警规则
前面在Prometheus server的配置文件中中指定了告警规则的文件为/etc/prometheus/rules.yml。
$ cat /etc/prometheus/rules.yml
groups:
- name: Warning
rules:
- alert: NodeMemoryUsage
expr: 100 - (node_memory_MemFree_bytes + node_memory_Cached_bytes + node_memory_Buffers_bytes) / node_memory_MemTotal_bytes*100 > 80
for: 1m
labels:
status: Warning
annotations:
summary: "{{$labels.instance}}: 内存使用率过高"
description: "{{$labels.instance}}: 内存使用率大于 80% (当前值: {{ $value }}"
- alert: NodeCpuUsage
expr: (1-((sum(increase(node_cpu_seconds_total{mode="idle"}[1m])) by (instance)) / (sum(increase(node_cpu_seconds_total[1m])) by (instance)))) * 100 > 70
for: 1m
labels:
status: Warning
annotations:
summary: "{{$labels.instance}}: CPU使用率过高"
description: "{{$labels.instance}}: CPU使用率大于 70% (当前值: {{ $value }}"
- alert: NodeDiskUsage
expr: 100 - node_filesystem_free_bytes{fstype=~"xfs|ext4"} / node_filesystem_size_bytes{fstype=~"xfs|ext4"} * 100 > 80
for: 1m
labels:
status: Warning
annotations:
summary: "{{$labels.instance}}: 分区使用率过高"
description: "{{$labels.instance}}: 分区使用大于 80% (当前值: {{ $value }}"
- alert: Node-UP
expr: up{job='node-exporter'} == 0
for: 1m
labels:
status: Warning
annotations:
summary: "{{$labels.instance}}: 服务宕机"
description: "{{$labels.instance}}: 服务中断超过1分钟"
- alert: TCP
expr: node_netstat_Tcp_CurrEstab > 1000
for: 1m
labels:
status: Warning
annotations:
summary: "{{$labels.instance}}: TCP连接过高"
description: "{{$labels.instance}}: 连接大于1000 (当前值: {{$value}})"
- alert: IO
expr: 100 - (avg(irate(node_disk_io_time_seconds_total[1m])) by(instance)* 100) < 60
for: 1m
labels:
status: Warning
annotations:
summary: "{{$labels.instance}}: 流入磁盘IO使用率过高"
description: "{{$labels.instance}}:流入磁盘IO大于60% (当前值:{{$value}})"
评论