前 言
Prometheus是一个用于监控和告警的开源系统。一开始由Soundcloud开发，后来在2016年，它迁移到CNCF并且成为Kubernetes之后最流行的项目之一。从整个Linux服务器到stand-alone web服务器、数据库服务或一个单独的进程，它都能监控。在Prometheus术语中，它所监控的事物称为目标（Target）。每个目标单元被称为指标（metric）。它以设置好的时间间隔通过http抓取目标，以收集指标并将数据放置在其时序数据库（Time Series Database）中。你可以使用PromQL查询语言查询相关target的指标。
本文中，我们将一步一步展示如何：
- 安装Prometheus（使用prometheus-operator Helm chart）以基于自定义事件进行监控/告警
- 创建和配置自定义告警规则，它将会在满足条件时发出告警
- 集成Alertmanager以处理由客户端应用程序（在本例中为Prometheus server）发送的告警
- 将Alertmanager与发送告警通知的邮件账户集成。
理解Prometheus及其抽象概念
从下图我们将看到所有组成Prometheus生态的组件：
Prometheus生态中的所有组件 （来源：http://prometheus.io）
以下是与本文相关的术语，大家可以快速了解：
- Prometheus Server：在时序数据库中抓取和存储指标的主要组件抓取：一种拉取方法以获取指标。它通常以10-60秒的时间间隔抓取。Target：检索数据的server客户端
- 服务发现：启用Prometheus，使其能够识别它需要监控的应用程序并在动态环境中拉取指标
- Alert Manager：负责处理警报的组件（包括silencing、inhibition、聚合告警信息，并通过邮件、PagerDuty、Slack等方式发送告警通知）。
- 数据可视化：抓取的数据存储在本地存储中，并使用PromQL直接查询，或通过Grafana dashboard查看。
理解Prometheus Operator
根据Prometheus Operator的项目所有者CoreOS称，Prometheus Operator可以配置原生Kubernetes并且可以管理和操作Prometheus和Alertmanager集群。
该Operator引入了以下Kubernetes自定义资源定义（CRDs）：Prometheus、ServiceMonitor、PrometheusRule和Alertmanager。如果你想了解更多内容可以访问链接：
https://github.com/coreos/prometheus-operator/blob/master/Documentation/design.md
在我们的演示中，我们将使用PrometheusRule来定义自定义规则。
首先，我们需要使用 stable/prometheus-operator Helm chart来安装Prometheus Operator，下载链接：
https://github.com/helm/charts/tree/master/stable/prometheus-operator
默认安装程序将会部署以下组件：prometheus-operator、prometheus、alertmanager、node-exporter、kube-state-metrics以及grafana。默认状态下，Prometheus将会抓取Kubernetes的主要组件：kube-apiserver、kube-controller-manager以及etcd。
安装Prometheus软件
前期准备
要顺利执行此次demo，你需要准备以下内容：
- 一个Google Cloud Platform账号（免费套餐即可）。其他任意云也可以
- Rancher v2.3.5（发布文章时的最新版本）
- 运行在GKE（版本1.15.9-gke.12.）上的Kubernetes集群（使用EKS或AKS也可以）
- 在计算机上安装好Helm binary
启动一个Rancher实例
直接按照这一直观的入门指南进行操作即可：
https://rancher.com/quick-start
使用Rancher部署一个GKE集群
使用Rancher来设置和配置你的Kubernetes集群：
https://rancher.com/docs/rancher/v2.x/en/cluster-provisioning/hosted-kubernetes-clusters/gke/
部署完成后，并且为kubeconfig文件配置了适当的credential和端点信息，就可以使用kubectl指向该特定集群。
部署Prometheus 软件
首先，检查一下我们所运行的Helm版本
<code>
$ kubectl create namespace monitoring
namespace/monitoring created/<code>
当我们使用Helm 3时，我们需要添加一个stable 镜像仓库，因为默认状态下不会设置该仓库。
<code>$ helm repo add stable https://kubernetes-charts.storage.googleapis.com
"stable" has been added to your repositories/<code>
<code>$ helm repo update
Hang tight while we grab the latest from your chart repositories...
...Successfully got an update from the "stable" chart repository
Update Complete. ⎈ Happy Helming!⎈/<code>
<code>$ helm repo list
NAME URL
stable https://kubernetes-charts.storage.googleapis.com/<code>
Helm配置完成后，我们可以开始安装prometheus-operator
<code>
$ kubectl create namespace monitoring
namespace/monitoring created/<code>
<code>
$ helm install --namespace monitoring demo stable/prometheus-operator
manifest_sorter.go:192: info: skipping unknown hook: "crd-install"
manifest_sorter.go:192: info: skipping unknown hook: "crd-install"
manifest_sorter.go:192: info: skipping unknown hook: "crd-install"
manifest_sorter.go:192: info: skipping unknown hook: "crd-install"
manifest_sorter.go:192: info: skipping unknown hook: "crd-install"
manifest_sorter.go:192: info: skipping unknown hook: "crd-install"
NAME: demo
LAST DEPLOYED: Sat Mar 14 09:40:35 2020
NAMESPACE: monitoring
STATUS: deployed
REVISION: 1
NOTES:
The Prometheus Operator has been installed. Check its status by running:
kubectl --namespace monitoring get pods -l "release=demo"
Visit https://github.com/coreos/prometheus-operator for instructions on how
to create & configure Alertmanager and Prometheus instances using the Operator./<code>
规 则
除了监控之外，Prometheus还让我们创建触发告警的规则。这些规则基于Prometheus的表达式语言。只要满足条件，就会触发告警并将其发送到Alertmanager。之后，我们会看到规则的具体形式。
我们回到demo。Helm完成部署之后，我们可以检查已经创建了什么pod：
<code>
$ kubectl -n monitoring get pods
NAME READY STATUS RESTARTS AGE
alertmanager-demo-prometheus-operator-alertmanager-0 2/2 Running 0 61s
demo-grafana-5576fbf669-9l57b 3/3 Running 0 72s
demo-kube-state-metrics-67bf64b7f4-4786k 1/1 Running 0 72s
demo-prometheus-node-exporter-ll8zx 1/1 Running 0 72s
demo-prometheus-node-exporter-nqnr6 1/1 Running 0 72s
demo-prometheus-node-exporter-sdndf 1/1 Running 0 72s
demo-prometheus-operator-operator-b9c9b5457-db9dj 2/2 Running 0 72s
prometheus-demo-prometheus-operator-prometheus-0 3/3 Running 1 50s/<code>
为了从web浏览器中访问Prometheus和Alertmanager，我们需要使用port转发。
由于本例中使用的是GCP实例，并且所有的kubectl命令都从该实例运行，因此我们使用实例的外部IP地址访问资源。
<code>
$ kubectl port-forward --address 0.0.0.0 -n monitoring prometheus-demo-prometheus-operator-prometheus-0 9090 >/dev/null 2>&1 &/<code>
<code>
$ kubectl port-forward --address 0.0.0.0 -n monitoring alertmanager-demo-prometheus-operator-alertmanager-0 9093 >/dev/null 2>&1 &/<code>
“Alert”选项卡向我们展示了所有当前正在运行/已配置的告警。也可以通过查询名称为prometheusrules的CRD从CLI进行检查：
<code>
$ kubectl -n monitoring get prometheusrules
NAME AGE
demo-prometheus-operator-alertmanager.rules 3m21s
demo-prometheus-operator-etcd 3m21s
demo-prometheus-operator-general.rules 3m21s
demo-prometheus-operator-k8s.rules 3m21s
demo-prometheus-operator-kube-apiserver-error 3m21s
demo-prometheus-operator-kube-apiserver.rules 3m21s
demo-prometheus-operator-kube-prometheus-node-recording.rules 3m21s
demo-prometheus-operator-kube-scheduler.rules 3m21s
demo-prometheus-operator-kubernetes-absent 3m21s
demo-prometheus-operator-kubernetes-apps 3m21s
demo-prometheus-operator-kubernetes-resources 3m21s
demo-prometheus-operator-kubernetes-storage 3m21s
demo-prometheus-operator-kubernetes-system 3m21s
demo-prometheus-operator-kubernetes-system-apiserver 3m21s
demo-prometheus-operator-kubernetes-system-controller-manager 3m21s
demo-prometheus-operator-kubernetes-system-kubelet 3m21s
demo-prometheus-operator-kubernetes-system-scheduler 3m21s
demo-prometheus-operator-node-exporter 3m21s
demo-prometheus-operator-node-exporter.rules 3m21s
demo-prometheus-operator-node-network 3m21s
demo-prometheus-operator-node-time 3m21s
demo-prometheus-operator-node.rules 3m21s
demo-prometheus-operator-prometheus 3m21s
demo-prometheus-operator-prometheus-operator 3m21s
/<code>
我们也可以检查位于prometheus容器中prometheus-operator Pod中的物理文件。
<code>$ kubectl -n monitoring exec -it prometheus-demo-prometheus-operator-prometheus-0 -- /bin/sh
Defaulting container name to prometheus.
Use 'kubectl describe pod/prometheus-demo-prometheus-operator-prometheus-0 -n monitoring' to see all of the containers in this pod./<code>
在容器中，我们可以检查规则的存储路径：
<code>/prometheus $ ls /etc/prometheus/rules/prometheus-demo-prometheus-operator-prometheus-rulefiles-0/
monitoring-demo-prometheus-operator-alertmanager.rules.yaml monitoring-demo-prometheus-operator-kubernetes-system-apiserver.yaml
monitoring-demo-prometheus-operator-etcd.yaml monitoring-demo-prometheus-operator-kubernetes-system-controller-manager.yaml
monitoring-demo-prometheus-operator-general.rules.yaml monitoring-demo-prometheus-operator-kubernetes-system-kubelet.yaml
monitoring-demo-prometheus-operator-k8s.rules.yaml monitoring-demo-prometheus-operator-kubernetes-system-scheduler.yaml
monitoring-demo-prometheus-operator-kube-apiserver-error.yaml monitoring-demo-prometheus-operator-kubernetes-system.yaml
monitoring-demo-prometheus-operator-kube-apiserver.rules.yaml monitoring-demo-prometheus-operator-node-exporter.rules.yaml
monitoring-demo-prometheus-operator-kube-prometheus-node-recording.rules.yaml monitoring-demo-prometheus-operator-node-exporter.yaml
monitoring-demo-prometheus-operator-kube-scheduler.rules.yaml monitoring-demo-prometheus-operator-node-network.yaml
monitoring-demo-prometheus-operator-kubernetes-absent.yaml monitoring-demo-prometheus-operator-node-time.yaml
monitoring-demo-prometheus-operator-kubernetes-apps.yaml monitoring-demo-prometheus-operator-node.rules.yaml
monitoring-demo-prometheus-operator-kubernetes-resources.yaml monitoring-demo-prometheus-operator-prometheus-operator.yaml
monitoring-demo-prometheus-operator-kubernetes-storage.yaml monitoring-demo-prometheus-operator-prometheus.yaml/<code>
为了详细了解如何将这些规则加载到Prometheus中，请检查Pod的详细信息。我们可以看到用于prometheus容器的配置文件是etc/prometheus/config_out/prometheus.env.yaml。该配置文件向我们展示了文件的位置或重新检查yaml的频率设置。
<code>
$ kubectl -n monitoring describe pod prometheus-demo-prometheus-operator-prometheus-0
/<code>
完整命令输出如下：
<code>Name: prometheus-demo-prometheus-operator-prometheus-0
Namespace: monitoring
Priority: 0
Node: gke-c-7dkls-default-0-c6ca178a-gmcq/10.132.0.15
Start Time: Wed, 11 Mar 2020 18:06:47 +0000
Labels: app=prometheus
controller-revision-hash=prometheus-demo-prometheus-operator-prometheus-5ccbbd8578
prometheus=demo-prometheus-operator-prometheus
statefulset.kubernetes.io/pod-name=prometheus-demo-prometheus-operator-prometheus-0
Annotations: <none>
Status: Running
IP: 10.40.0.7
IPs: <none>
Controlled By: StatefulSet/prometheus-demo-prometheus-operator-prometheus
Containers:
prometheus:
Container ID: docker://360db8a9f1cce8d72edd81fcdf8c03fe75992e6c2c59198b89807aa0ce03454c
Image: quay.io/prometheus/prometheus:v2.15.2
Image ID: docker-pullable://quay.io/prometheus/prometheus@sha256:914525123cf76a15a6aaeac069fcb445ce8fb125113d1bc5b15854bc1e8b6353
Port: 9090/TCP
Host Port: 0/TCP
Args:
--web.console.templates=/etc/prometheus/consoles
--web.console.libraries=/etc/prometheus/console_libraries
--config.file=/etc/prometheus/config_out/prometheus.env.yaml
--storage.tsdb.path=/prometheus
--storage.tsdb.retention.time=10d
--web.enable-lifecycle
--storage.tsdb.no-lockfile
--web.external-url=http://demo-prometheus-operator-prometheus.monitoring:9090
--web.route-prefix=/
State: Running
Started: Wed, 11 Mar 2020 18:07:07 +0000
Last State: Terminated
Reason: Error
Message: caller=main.go:648 msg="Starting TSDB ..."
level=info ts=2020-03-11T18:07:02.185Z caller=web.go:506 component=web msg="Start listening for connections" address=0.0.0.0:9090
level=info ts=2020-03-11T18:07:02.192Z caller=head.go:584 component=tsdb msg="replaying WAL, this may take awhile"
level=info ts=2020-03-11T18:07:02.192Z caller=head.go:632 component=tsdb msg="WAL segment loaded" segment=0 maxSegment=0
level=info ts=2020-03-11T18:07:02.194Z caller=main.go:663 fs_type=EXT4_SUPER_MAGIC
level=info ts=2020-03-11T18:07:02.194Z caller=main.go:664 msg="TSDB started"
level=info ts=2020-03-11T18:07:02.194Z caller=main.go:734 msg="Loading configuration file" filename=/etc/prometheus/config_out/prometheus.env.yaml
level=info ts=2020-03-11T18:07:02.194Z caller=main.go:517 msg="Stopping scrape discovery manager..."
level=info ts=2020-03-11T18:07:02.194Z caller=main.go:531 msg="Stopping notify discovery manager..."
level=info ts=2020-03-11T18:07:02.194Z caller=main.go:553 msg="Stopping scrape manager..."
level=info ts=2020-03-11T18:07:02.194Z caller=manager.go:814 component="rule manager" msg="Stopping rule manager..."
level=info ts=2020-03-11T18:07:02.194Z caller=manager.go:820 component="rule manager" msg="Rule manager stopped"
level=info ts=2020-03-11T18:07:02.194Z caller=main.go:513 msg="Scrape discovery manager stopped"
level=info ts=2020-03-11T18:07:02.194Z caller=main.go:527 msg="Notify discovery manager stopped"
level=info ts=2020-03-11T18:07:02.194Z caller=main.go:547 msg="Scrape manager stopped"
level=info ts=2020-03-11T18:07:02.197Z caller=notifier.go:598 component=notifier msg="Stopping notification manager..."
level=info ts=2020-03-11T18:07:02.197Z caller=main.go:718 msg="Notifier manager stopped"
level=error ts=2020-03-11T18:07:02.197Z caller=main.go:727 err="error loading config from \"/etc/prometheus/config_out/prometheus.env.yaml\": couldn't load configuration (--config.file=\"/etc/prometheus/config_out/prometheus.env.yaml\"): open /etc/prometheus/config_out/prometheus.env.yaml: no such file or directory"
Exit Code: 1
Started: Wed, 11 Mar 2020 18:07:02 +0000
Finished: Wed, 11 Mar 2020 18:07:02 +0000
Ready: True
Restart Count: 1
Liveness: http-get http://:web/-/healthy delay=0s timeout=3s period=5s #success=1 #failure=6
Readiness: http-get http://:web/-/ready delay=0s timeout=3s period=5s #success=1 #failure=120
Environment: <none>
Mounts:
/etc/prometheus/certs from tls-assets (ro)
/etc/prometheus/config_out from config-out (ro)
/etc/prometheus/rules/prometheus-demo-prometheus-operator-prometheus-rulefiles-0 from prometheus-demo-prometheus-operator-prometheus-rulefiles-0 (rw)
/prometheus from prometheus-demo-prometheus-operator-prometheus-db (rw)
/var/run/secrets/kubernetes.io/serviceaccount from demo-prometheus-operator-prometheus-token-jvbrr (ro)
prometheus-config-reloader:
Container ID: docker://de27cdad7067ebd5154c61b918401b2544299c161850daf3e317311d2d17af3d
Image: quay.io/coreos/prometheus-config-reloader:v0.37.0
Image ID: docker-pullable://quay.io/coreos/prometheus-config-reloader@sha256:5e870e7a99d55a5ccf086063efd3263445a63732bc4c04b05cf8b664f4d0246e
Port: <none>
Host Port: <none>
Command:
/bin/prometheus-config-reloader
Args:
--log-format=logfmt
--reload-url=http://127.0.0.1:9090/-/reload
--config-file=/etc/prometheus/config/prometheus.yaml.gz
--config-envsubst-file=/etc/prometheus/config_out/prometheus.env.yaml
State: Running
Started: Wed, 11 Mar 2020 18:07:04 +0000
Ready: True
Restart Count: 0
Limits:
cpu: 100m
memory: 25Mi
Requests:
cpu: 100m
memory: 25Mi
Environment:
POD_NAME: prometheus-demo-prometheus-operator-prometheus-0 (v1:metadata.name)
Mounts:
/etc/prometheus/config from config (rw)
/etc/prometheus/config_out from config-out (rw)
/var/run/secrets/kubernetes.io/serviceaccount from demo-prometheus-operator-prometheus-token-jvbrr (ro)
rules-configmap-reloader:
Container ID: docker://5804e45380ed1b5374a4c2c9ee4c9c4e365bee93b9ccd8b5a21f50886ea81a91
Image: quay.io/coreos/configmap-reload:v0.0.1
Image ID: docker-pullable://quay.io/coreos/configmap-reload@sha256:e2fd60ff0ae4500a75b80ebaa30e0e7deba9ad107833e8ca53f0047c42c5a057
Port: <none>
Host Port: <none>
Args:
--webhook-url=http://127.0.0.1:9090/-/reload
--volume-dir=/etc/prometheus/rules/prometheus-demo-prometheus-operator-prometheus-rulefiles-0
State: Running
Started: Wed, 11 Mar 2020 18:07:06 +0000
Ready: True
Restart Count: 0
Limits:
cpu: 100m
memory: 25Mi
Requests:
cpu: 100m
memory: 25Mi
Environment: <none>
Mounts:
/etc/prometheus/rules/prometheus-demo-prometheus-operator-prometheus-rulefiles-0 from prometheus-demo-prometheus-operator-prometheus-rulefiles-0 (rw)
/var/run/secrets/kubernetes.io/serviceaccount from demo-prometheus-operator-prometheus-token-jvbrr (ro)
Conditions:
Type Status
Initialized True
Ready True
ContainersReady True
PodScheduled True
Volumes:
config:
Type: Secret (a volume populated by a Secret)
SecretName: prometheus-demo-prometheus-operator-prometheus
Optional: false
tls-assets:
Type: Secret (a volume populated by a Secret)
SecretName: prometheus-demo-prometheus-operator-prometheus-tls-assets
Optional: false
config-out:
Type: EmptyDir (a temporary directory that shares a pod's lifetime)
Medium:
SizeLimit: <unset>
prometheus-demo-prometheus-operator-prometheus-rulefiles-0:
Type: ConfigMap (a volume populated by a ConfigMap)
Name: prometheus-demo-prometheus-operator-prometheus-rulefiles-0
Optional: false
prometheus-demo-prometheus-operator-prometheus-db:
Type: EmptyDir (a temporary directory that shares a pod's lifetime)
Medium:
SizeLimit: <unset>
demo-prometheus-operator-prometheus-token-jvbrr:
Type: Secret (a volume populated by a Secret)
SecretName: demo-prometheus-operator-prometheus-token-jvbrr
Optional: false
QoS Class: Burstable
Node-Selectors: <none>
Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s
node.kubernetes.io/unreachable:NoExecute for 300s
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled 4m51s default-scheduler Successfully assigned monitoring/prometheus-demo-prometheus-operator-prometheus-0 to gke-c-7dkls-default-0-c6ca178a-gmcq
Normal Pulling 4m45s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Pulling image "quay.io/prometheus/prometheus:v2.15.2"
Normal Pulled 4m39s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Successfully pulled image "quay.io/prometheus/prometheus:v2.15.2"
Normal Pulling 4m36s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Pulling image "quay.io/coreos/prometheus-config-reloader:v0.37.0"
Normal Pulled 4m35s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Successfully pulled image "quay.io/coreos/prometheus-config-reloader:v0.37.0"
Normal Pulling 4m34s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Pulling image "quay.io/coreos/configmap-reload:v0.0.1"
Normal Started 4m34s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Started container prometheus-config-reloader
Normal Created 4m34s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Created container prometheus-config-reloader
Normal Pulled 4m33s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Successfully pulled image "quay.io/coreos/configmap-reload:v0.0.1"
Normal Created 4m32s (x2 over 4m36s) kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Created container prometheus
Normal Created 4m32s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Created container rules-configmap-reloader
Normal Started 4m32s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Started container rules-configmap-reloader
Normal Pulled 4m32s kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Container image "quay.io/prometheus/prometheus:v2.15.2" already present on machine
Normal Started 4m31s (x2 over 4m36s) kubelet, gke-c-7dkls-default-0-c6ca178a-gmcq Started container prometheus/<none>/<unset>/<unset>/<none>/<none>/<none>/<none>/<none>/<none>/<none>/<none>/<code>
让我们清理默认规则，使得我们可以更好地观察我们将要创建的那个规则。以下命令将删除所有规则，但会留下monitoring-demo-prometheus-operator-alertmanager.rules。
<code>$ kubectl -n monitoring delete prometheusrules $(kubectl -n monitoring get prometheusrules | grep -v alert)/<code>
<code>
$ kubectl -n monitoring get prometheusrules
NAME AGE
demo-prometheus-operator-alertmanager.rules 8m53s
/<code>
请注意：我们只保留一条规则是为了让demo更容易。但是有一条规则，你绝对不能删除，它位于monitoring-demo-prometheus-operator-general.rules.yaml中，被称为看门狗。该告警总是处于触发状态，其目的是确保整个告警流水线正常运转。
让我们从CLI中检查我们留下的规则并将其与我们将在浏览器中看到的进行比较。
<code>$ kubectl -n monitoring describe prometheusrule demo-prometheus-operator-alertmanager.rules
Name: demo-prometheus-operator-alertmanager.rules
Namespace: monitoring
Labels: app=prometheus-operator
chart=prometheus-operator-8.12.1
heritage=Tiller
release=demo
Annotations: prometheus-operator-validated: true
API Version: monitoring.coreos.com/v1
Kind: PrometheusRule
Metadata:
Creation Timestamp: 2020-03-11T18:06:25Z
Generation: 1
Resource Version: 4871
Self Link: /apis/monitoring.coreos.com/v1/namespaces/monitoring/prometheusrules/demo-prometheus-operator-alertmanager.rules
UID: 6a84dbb0-feba-4f17-b3dc-4b6486818bc0
Spec:
Groups:
Name: alertmanager.rules
Rules:
Alert: AlertmanagerConfigInconsistent
Annotations:
Message: The configuration of the instances of the Alertmanager cluster `{{$labels.service}}` are out of sync.
Expr: count_values("config_hash", alertmanager_config_hash{job="demo-prometheus-operator-alertmanager",namespace="monitoring"}) BY (service) / ON(service) GROUP_LEFT() label_replace(max(prometheus_operator_spec_replicas{job="demo-prometheus-operator-operator",namespace="monitoring",controller="alertmanager"}) by (name, job, namespace, controller), "service", "$1", "name", "(.*)") != 1
For: 5m
Labels:
Severity: critical
Alert: AlertmanagerFailedReload
Annotations:
Message: Reloading Alertmanager's configuration has failed for {{ $labels.namespace }}/{{ $labels.pod}}.
Expr: alertmanager_config_last_reload_successful{job="demo-prometheus-operator-alertmanager",namespace="monitoring"} == 0
For: 10m
Labels:
Severity: warning
Alert: AlertmanagerMembersInconsistent
Annotations:
Message: Alertmanager has not found all other members of the cluster.
Expr: alertmanager_cluster_members{job="demo-prometheus-operator-alertmanager",namespace="monitoring"}
!= on (service) GROUP_LEFT()
count by (service) (alertmanager_cluster_members{job="demo-prometheus-operator-alertmanager",namespace="monitoring"})
For: 5m
Labels:
Severity: critical
Events: <none>/<code>
让我们移除所有默认告警并创建一个我们自己的告警：
<code>$ kubectl -n monitoring edit prometheusrules demo-prometheus-operator-alertmanager.rules
prometheusrule.monitoring.coreos.com/demo-prometheus-operator-alertmanager.rules edited
/<code>
我们的自定义告警如下所示：
<code>
$ kubectl -n monitoring describe prometheusrule demo-prometheus-operator-alertmanager.rules
Name: demo-prometheus-operator-alertmanager.rules
Namespace: monitoring
Labels: app=prometheus-operator
chart=prometheus-operator-8.12.1
heritage=Tiller
release=demo
Annotations: prometheus-operator-validated: true
API Version: monitoring.coreos.com/v1
Kind: PrometheusRule
Metadata:
Creation Timestamp: 2020-03-11T18:06:25Z
Generation: 3
Resource Version: 18180
Self Link: /apis/monitoring.coreos.com/v1/namespaces/monitoring/prometheusrules/demo-prometheus-operator-alertmanager.rules
UID: 6a84dbb0-feba-4f17-b3dc-4b6486818bc0
Spec:
Groups:
Name: alertmanager.rules
Rules:
Alert: PodHighCpuLoad
Annotations:
Message: Alertmanager has found {{ $labels.instance }} with CPU too high
Expr: rate (container_cpu_usage_seconds_total{pod_name=~"nginx-.*", image!="", container!="POD"}[5m]) > 0.04
For: 1m
Labels:
Severity: critical
Events: <none>/<code>
以下是我们创建的告警的选项：
- annotation：描述告警的信息标签集。
- expr：由PromQL写的表达式
- for：可选参数，设置了之后会告诉Prometheus在定义的时间段内告警是否处于active状态。仅在此定义时间后才会触发告警。
- label：可以附加到告警的额外标签。如果你想了解更多关于告警的信息，可以访问：https://prometheus.io/docs/prometheus/latest/configuration/alerting_rules/
现在我们已经完成了Prometheus告警的设置，让我们配置Alertmanager，使得我们能够通过电子邮件获得告警通知。Alertmanager的配置位于Kubernetes secret对象中。
<code>
$ kubectl get secrets -n monitoring
NAME TYPE DATA AGE
alertmanager-demo-prometheus-operator-alertmanager Opaque 1 32m
default-token-x4rgq kubernetes.io/service-account-token 3 37m
demo-grafana Opaque 3 32m
demo-grafana-test-token-p6qnk kubernetes.io/service-account-token 3 32m
demo-grafana-token-ff6nl kubernetes.io/service-account-token 3 32m
demo-kube-state-metrics-token-vmvbr kubernetes.io/service-account-token 3 32m
demo-prometheus-node-exporter-token-wlnk9 kubernetes.io/service-account-token 3 32m
demo-prometheus-operator-admission Opaque 3 32m
demo-prometheus-operator-alertmanager-token-rrx4k kubernetes.io/service-account-token 3 32m
demo-prometheus-operator-operator-token-q9744 kubernetes.io/service-account-token 3 32m
demo-prometheus-operator-prometheus-token-jvbrr kubernetes.io/service-account-token 3 32m
prometheus-demo-prometheus-operator-prometheus Opaque 1 31m
prometheus-demo-prometheus-operator-prometheus-tls-assets Opaque 0 31m
/<code>
我们只对alertmanager-demo-prometheus-operator-alertmanager感兴趣。让我们看一下：
<code>
kubectl -n monitoring get secret alertmanager-demo-prometheus-operator-alertmanager -o yaml
apiVersion: v1
data:
alertmanager.yaml: Z2xvYmFsOgogIHJlc29sdmVfdGltZW91dDogNW0KcmVjZWl2ZXJzOgotIG5hbWU6ICJudWxsIgpyb3V0ZToKICBncm91cF9ieToKICAtIGpvYgogIGdyb3VwX2ludGVydmFsOiA1bQogIGdyb3VwX3dhaXQ6IDMwcwogIHJlY2VpdmVyOiAibnVsbCIKICByZXBlYXRfaW50ZXJ2YWw6IDEyaAogIHJvdXRlczoKICAtIG1hdGNoOgogICAgICBhbGVydG5hbWU6IFdhdGNoZG9nCiAgICByZWNlaXZlcjogIm51bGwiCg==
kind: Secret
metadata:
creationTimestamp: "2020-03-11T18:06:24Z"
labels:
app: prometheus-operator-alertmanager
chart: prometheus-operator-8.12.1
heritage: Tiller
release: demo
name: alertmanager-demo-prometheus-operator-alertmanager
namespace: monitoring
resourceVersion: "3018"
selfLink: /api/v1/namespaces/monitoring/secrets/alertmanager-demo-prometheus-operator-alertmanager
uid: 6baf6883-f690-47a1-bb49-491935956c22
type: Opaque/<code>
alertmanager.yaml字段是由base64编码的，让我们看看：
<code>
$ echo 'Z2xvYmFsOgogIHJlc29sdmVfdGltZW91dDogNW0KcmVjZWl2ZXJzOgotIG5hbWU6ICJudWxsIgpyb3V0ZToKICBncm91cF9ieToKICAtIGpvYgogIGdyb3VwX2ludGVydmFsOiA1bQogIGdyb3VwX3dhaXQ6IDMwcwogIHJlY2VpdmVyOiAibnVsbCIKICByZXBlYXRfaW50ZXJ2YWw6IDEyaAogIHJvdXRlczoKICAtIG1hdGNoOgogICAgICBhbGVydG5hbWU6IFdhdGNoZG9nCiAgICByZWNlaXZlcjogIm51bGwiCg==' | base64 --decode
global:
resolve_timeout: 5m
receivers:
- name: "null"
route:
group_by:
- job
group_interval: 5m
group_wait: 30s
receiver: "null"
repeat_interval: 12h
routes:
- match:
alertname: Watchdog
receiver: "null"/<code>
正如我们所看到的，这是默认的Alertmanager配置。你也可以在Alertmanager UI的Status选项卡中查看此配置。接下来，我们来对它进行一些更改——在本例中为发送邮件：
<code>
$ cat alertmanager.yaml
global:
resolve_timeout: 5m
route:
group_by: [Alertname]
# Send all notifications to me.
receiver: demo-alert
group_wait: 30s
group_interval: 5m
repeat_interval: 12h
routes:
- match:
alertname: DemoAlertName
receiver: 'demo-alert'
receivers:
- name: demo-alert
email_configs:
- to: [email protected]
from: [email protected]
# Your smtp server address
smarthost: smtp.gmail.com:587
auth_username: [email protected]
auth_identity: [email protected]
auth_password: 16letter_generated token # you can use gmail account password, but better create a dedicated token for this
headers:
From: [email protected]
Subject: 'Demo ALERT'/<code>
首先，我们需要对此进行编码：
<code>
$ cat alertmanager.yaml | base64 -w0/<code>
我们获得编码输出后，我们需要在我们将要应用的yaml文件中填写它：
<code>
cat alertmanager-secret-k8s.yaml
apiVersion: v1
data:
alertmanager.yaml: <paste>
kind: Secret
metadata:
name: alertmanager-demo-prometheus-operator-alertmanager
namespace: monitoring
type: Opaque/<paste>/<code>
<code>$ kubectl apply -f alertmanager-secret-k8s.yaml
Warning: kubectl apply should be used on resource created by either kubectl create --save-config or kubectl apply
secret/alertmanager-demo-prometheus-operator-alertmanager configured/<code>
该配置将会自动重新加载并在UI中显示更改。
接下来，我们部署一些东西来对其进行监控。对于本例而言，一个简单的nginx deployment已经足够：
<code>
$ cat nginx-deployment.yaml
apiVersion: apps/v1 # for versions before 1.9.0 use apps/v1beta2
kind: Deployment
metadata:
name: nginx-deployment
spec:
selector:
matchLabels:
app: nginx
replicas: 3 # tells deployment to run 2 pods matching the template
template:
metadata:
labels:
app: nginx
spec:
containers:
- name: nginx
image: nginx:1.7.9
ports:
- containerPort: 80/<code>
<code>
$ kubectl apply -f nginx-deployment.yaml
deployment.apps/nginx-deployment created/<code>
根据配置yaml，我们有3个副本：
<code>
$ kubectl get pods
NAME READY STATUS RESTARTS AGE
nginx-deployment-5754944d6c-7g6gq 1/1 Running 0 67s
nginx-deployment-5754944d6c-lhvx8 1/1 Running 0 67s
nginx-deployment-5754944d6c-whhtr 1/1 Running 0 67s/<code>
在Prometheus UI中，使用我们为告警配置的相同表达式：
<code>
rate (container_cpu_usage_seconds_total{pod_name=~"nginx-.*", image!="", container!="POD"}[5m])/<code>
我们可以为这些Pod检查数据，所有Pod的值应该为0。
让我们在其中一个pod中添加一些负载，然后来看看值的变化，当值大于0.04时，我们应该接收到告警：
<code>
$ kubectl exec -it nginx-deployment-5754944d6c-7g6gq -- /bin/sh
# yes > /dev/null
/<code>
该告警有3个阶段：
- Inactive：不满足告警触发条件
- Pending：条件已满足
- Firing：触发告警
我们已经看到告警处于inactive状态，所以在CPU上添加一些负载，以观察到剩余两种状态：
告警一旦触发，将会在Alertmanager中显示：
Alertmanger配置为当我们收到告警时发送邮件。所以此时，如果我们检查收件箱，会看到类似以下内容：
总 结
我们知道监控的重要性，但是如果没有告警，它将是不完整的。发生问题时，告警可以立即通知我们，让我们立即了解到系统出现了问题。而Prometheus涵盖了这两个方面：既有监控解决方案又通过Alertmanager组件发出告警。本文中，我们看到了如何在Prometheus配置中定义告警以及告警在触发时如何到达Alertmanager。然后根据Alertmanager的定义/集成，我们收到了一封电子邮件，其中包含触发的告警的详细信息（也可以通过Slack或PagerDuty发送）。
分享到:
閱讀更多 RancherLabs 的文章
