为Kubernetes集群部署一个ChatGPT机器人

频道：行业资讯日期：2023-04-01 20:36:05 浏览：1061

今天我想分享一个有趣的项目叫做 “K8s ChatGPT Bot[1]”。该项目的目的是为 K8s 集群部署一个 ChatGPT 机器人。我们可以询问 ChatGPT 帮助我们解决 Prometheus 的警报，可以得到简洁的答复，不再需要一人在黑暗中独自解决警报问题！

我们需要用到 Robusta[2]，如果你还没有 Robusta，可以参考《K8s — Robusta, K8s Troubleshooting Platform[3]》搭建一个 Robusta 平台。

下图是 Robusta 平台如何工作的截图：

你可以在此处查看完整的演示视频：

https://www.loom.com/share/964cd8735a874287a9155c77320bdcdb

—1—

运行 K8s ChatGPT 机器人项目

该机器人项目是基于 Robusta.dev[4] 实现的，Robusta.dev 是一个用于响应 K8s 警报的开源平台。其工作流程大致如下：

Prometheus 使用 Webhook 接收器将警报转发给 Robusta.dev 。Robusta.dev 询问 ChatGPT 如何修复 Prometheus 警报。

—2—

先决条件

SlackKubernetes集群Python 3.7 及以上

—3—

如何安装 Robusta

生成 Robusta 配置文件

为 Robusta 准备 Python 虚拟环境。

$python3.10-mvenvrobusta $sourcerobusta/bin/activate (robusta)$pipinstall-Urobusta-cli--no-cache Collectingrobusta-cli Downloadingrobusta_cli-0.10.10-py3-none-any.whl(223kB) ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━223.8/223.8kB30.0MB/seta0:00:00 Collectingpymsteams<0.2.0,>=0.1.16 Downloadingpymsteams-0.1.16.tar.gz(7.6kB) Preparingmetadata(setup.py)...done ... SuccessfullyinstalledPyJWT-2.4.0appdirs-1.4.4autopep8-2.0.1black-21.5b2 cachetools-5.2.1certifi-2022.12.7cffi-1.15.1charset-normalizer-3.0.1 ... ruamel.yaml.clib-0.2.7six-1.16.0slack-sdk-3.19.5tenacity-8.1.0 toml-0.10.2tomli-2.0.1typer-0.4.2typing-extensions-4.4.0urllib3-1.26.14 watchgod-0.7webexteamssdk-1.6.1websocket-client-1.3.3

使用 robusta 生成一个配置文件：

配置 Slack 集成

使用浏览器打开网页：https://api.robusta.dev/integrations/slack?id=xxxx

更新权限：

恭喜你配置 Slack 集成成功。

现在回到我们的 Terminal 终端，我们可以看到以下内容，说明操作成功：

$robustagen-config Robustareportsitsfindingstoexternaldestinations(wecallthem"sinks"). Welldefinesomeofthemnow. ConfigureSlackintegration?ThisisHIGHLYrecommended.[Y/n]:Y Ifyourbrowserdoesnotautomaticallylaunch,openthebelowurl: https://api.robusta.dev/integrations/slack?id=xxxx YouvejustconnectedRobustatotheSlackof:devopsfans WhichslackchannelshouldIsendnotificationsto?#k8s-chatgpt-bot ConfigureRobustaUIsink?ThisisHIGHLYrecommended.[Y/n]:Y EnteryourGmail/Googleaddress.Thiswillbeusedtologin:xxx@gmail.com Chooseyouraccountname(e.gyourorganizationname):devopsfans Successfullyregistered. RobustacanusePrometheusasanalertsource. Ifyouhaventinstalledityet,Robustacaninstalla pre-configuredPrometheus. Wouldyouliketodoso?[y/N]:y PleasereadandapproveourEndUserLicenseAgreement: https://api.robusta.dev/eula.html DoyouacceptourEndUserLicenseAgreement?[y/N]:y Lastquestion!WouldyouliketohelpusimproveRobustabysendingexceptionreports?[y/N]:N Savedconfigurationto./generated_values.yaml-savethisfileforfutureuse! FinishinstallingwithHelm(seetheRobustadocs). ThenlogintoRobustaUIathttps://platform.robusta.dev Bytheway,wellsendyousomemessageslatertogetfeedback. (WedontstoreyourAPIkey,sowescheduledfuturemessagesusingSlacks API)

在 slack channel 中，我们还可以看到：

使用 Helm3 安装 Robusta

安装和更新 robusta 仓库。

$helmrepoaddrobustahttps://robusta-charts.storage.googleapis.com&&helmrepoupdate "robusta"hasbeenaddedtoyourrepositories Hangtightwhilewegrabthelatestfromyourchartrepositories... ...Successfullygotanupdatefromthe"kedacore"chartrepository ...Successfullygotanupdatefromthe"robusta"chartrepository ...Successfullygotanupdatefromthe"grafana"chartrepository ...Successfullygotanupdatefromthe"prometheus-community"chartrepository ...Successfullygotanupdatefromthe"stable"chartrepository UpdateComplete.⎈HappyHelming!⎈

更新 generated_values.yaml 文件

使用以下内容更新 generated_values.yaml 文件：

playbookRepos: chatgpt_robusta_actions: url:"https://github.com/robusta-dev/kubernetes-chatgpt-bot.git" customPlaybooks: #AddtheAskChatGPTbuttontoallPrometheusalerts -triggers: -on_prometheus_alert:{} actions: -chat_gpt_enricher:{} globalConfig: chat_gpt_token:YOURKEYGOESHERE

将Robusta 部署到 K8s

$helminstallrobustarobusta/robusta-f./generated_values.yaml\ --setclusterName=dev-cluster

验证两个 Robusta pod 正常运行且在日志中没有发现错误日志：

$kubectlgetpods-A|greprobusta defaultalertmanager-robusta-kube-prometheus-st-alertmanager-02/2Running1(4m19sago)9m25s defaultprometheus-robusta-kube-prometheus-st-prometheus-02/2Running09m25s defaultrobusta-forwarder-6b7d8d9d88-2rv9d1/1Running09m29s defaultrobusta-grafana-64944bfcdc-v97xh3/3Running09m29s defaultrobusta-kube-prometheus-st-admission-patch-6zj4b0/1Completed09m28s defaultrobusta-kube-prometheus-st-operator-7b985d7fb-c9f9t1/1Running09m29s defaultrobusta-kube-state-metrics-688d794968-ll6gf1/1Running09m29s defaultrobusta-prometheus-node-exporter-2k5f71/1Running05m24s defaultrobusta-prometheus-node-exporter-zxsrg1/1Running09m29s defaultrobusta-runner-5868b494d6-m62921/1Running09m29s $robustalogs settingupcoloredlogging 2023-01-1422:57:01.428INFOloggerinitializedusingINFOloglevel 2023-01-1422:57:01.429INFOCreatinghikarumonkeypatches 2023-01-1422:57:01.429INFOCreatingyamlmonkeypatch 2023-01-1422:57:01.429INFOCreatingkubernetesContainerImagemonkeypatch 2023-01-1422:57:01.430INFOwatchingdir/etc/robusta/playbooks/forcustomplaybookschanges 2023-01-1422:57:01.431INFOwatchingdir/etc/robusta/config/active_playbooks.yamlforcustomplaybookschanges 2023-01-1422:57:01.431INFOReloadingplaybookpackagesduetochangeoninitialization 2023-01-1422:57:01.431INFOloadingconfig/etc/robusta/config/active_playbooks.yaml 2023-01-1422:57:01.467INFONocustomplaybooksdefinedat/etc/robusta/playbooks/storage 2023-01-1422:57:01.468INFOCloninggitrepohttps://github.com/robusta-dev/kubernetes-chatgpt-bot.git.reponamekubernetes-chatgpt-bot ... 2023-01-1422:57:07.364INFOconnectingtoserverasaccount_id=8302df56-c554-4129-8b95-d143d1f2e3a2;cluster_name=dev-cluster 2023-01-1422:57:07.977INFOInitializingservicescache 2023-01-1422:57:08.203INFOInitializingnodescache 2023-01-1422:57:08.395INFOInitializingjobscache 2023-01-1422:57:08.603INFOGettingeventshistory 2023-01-1422:57:10.403INFOClusterhistoricaldatasent. 2023-01-1423:04:43.681INFOclusterstatus{account_id:8302df56-c554-4129-8b95-d143d1f2e3a2,cluster_id:dev-cluster,version:0.10.10,last_alert_at:2023-01-1423:04:18.959377,light_actions:[related_pods,prometheus_enricher,add_silence,delete_pod,delete_silence,get_silences,logs_enricher,pod_events_enricher,deployment_events_enricher,job_events_enricher,job_pod_enricher,get_resource_yaml,node_cpu_enricher,node_disk_analyzer,node_running_pods_enricher,node_allocatable_resources_enricher,node_status_enricher,node_graph_enricher,oomkilled_container_graph_enricher,pod_oom_killer_enricher,oom_killer_enricher,volume_analysis,python_profiler,pod_ps,python_memory,debugger_stack_trace,python_process_inspector,prometheus_alert,create_pvc_snapshot],updated_at:now()}

—4—

使用 Robusta

现在，我们终于可以使用 Robusta 了！默认情况下，Robusta 在当 K8s Pod 崩溃时发送通知。

因此，让我们创建一个崩溃的 Pod：

$kubectlapply-fhttps://gist.githubusercontent.com/robusta-lab/283609047306dc1f05cf59806ade30b6/raw deployment.apps/crashpodcreated $kubectlgetpods-A|grepcrash defaultcrashpod-64db77b594-cgz4s0/1CrashLoopBackOff2(21sago)36s

一旦 Pod 达到两次重启，在 Slack channel 我们就将收到有关 Pod 崩溃的消息，如下所示：

—5—

与 ChatGPT 互动

经过我们的实验，我们已经确认 Robusta 已与我们的 Slack 和 K8s 集群集成。接下来让我们与 ChatGPT 机器人进行交互！

立即触发 Prometheus 警报，跳过正常延迟：

$robustaplaybookstriggerprometheus_alertalert_name=KubePodCrashLoopingnamespace=defaultpod_name=example-pod ====================================================================== Triggeringaction... ====================================================================== runningcmd:curl-XPOSThttp://localhost:5000/api/trigger-HContent-Type:application/json-d {"action_name":"prometheus_alert","action_params": {"alert_name":"KubePodCrashLooping","namespace":"default", "pod_name":"example-pod"}} {"success":true} ====================================================================== Fetchinglogs... ====================================================================== 2023-01-1423:14:33.463INFOErrorloadingkubernetespoddefault/example-pod.reason:NotFoundstatus:404 2023-01-1423:14:33.481INFOErrorloadingkubernetespoddefault/example-pod.reason:NotFoundstatus:404 2023-01-1423:14:33.503INFOErrorloadingkubernetespoddefault/example-pod.reason:NotFoundstatus:404 2023-01-1423:14:33.505ERRORcannotrunpod_events_enricheronalertwithnopodobject:PrometheusKubernetesAlert(sink_findings=defaultdict(<classlist>,{main_slack_sink:[<robusta.core.reporting.base.Findingobjectat0x7fab53074e20>],main_ms_teams_sink:[<robusta.core.reporting.base.Findingobjectat0x7fab53074700>],robusta_ui_sink:[<robusta.core.reporting.base.Findingobjectat0x7fab40773a30>]}),named_sinks=[main_slack_sink,main_ms_teams_sink,robusta_ui_sink],response={success:True},stop_processing=False,_scheduler=<robusta.integrations.scheduled.playbook_scheduler_manager_impl.PlaybooksSchedulerManagerImplobjectat0x7fab4088e0a0>,_context=ExecutionContext(account_id=8302df56-c554-4129-8b95-d143d1f2e3a2,cluster_name=dev-cluster),obj=None,alert=PrometheusAlert(endsAt=datetime.datetime(2023,1,14,23,14,33,430401),generatorURL=,startsAt=datetime.datetime(2023,1,14,23,14,33,430406),fingerprint=,status=firing,labels={severity:error,namespace:default,alertname:KubePodCrashLooping,pod:example-pod},annotations={}),alert_name=KubePodCrashLooping,alert_severity=error,label_namespace=default,node=None,pod=None,deployment=None,job=None,daemonset=None,statefulset=None) 2023-01-1423:14:33.524INFOErrorloadingkubernetespoddefault/example-pod.reason:NotFoundstatus:404 2023-01-1423:14:33.696ERRORCallbackBlocknotsupportedformsteams 2023-01-1423:14:33.697ERRORerrorsendingmessagetomsteams e=InvalidURLFalse:Noschemasupplied.Perhapsyoumeanthttp://False? ====================================================================== Done! ======================================================================

现在切换到 Slack，我们将看到一个新警报，并且这次带有 “Ask ChatGPT” 的按钮！

就是这样！恭喜，我们刚刚成功安装了我们的第一个 K8s ChatGPT 机器人！

示例 2：节点容量为 100%

下面是一个节点达到 100% 容量的例子：

—6—

Robusta UI

Robusta 有一个用于集成的 UI，也有一个预配置的 Promethus 系统，如果你还没有自己的 K8s 集群，并且想尝试一下这个 ChatGPT 机器人，你可以使用 Robusta 现有的！