• DevOps

  • devops流水线未运行

zhangzl419 如果执行了上面3步，还不能运行是流水线，麻烦在重启下 devops-controller 服务试下，如果还不行，看下 devops-controller 日志有没有相关报错信息；

      7 天 后

      yudong

      碰到类似的问题：
      1，开始发现流水线 “未运行”，查看devops-jenkins日志：

      023-08-21 03:07:25.422+0000 [id=20295] WARNING i.k.j.d.a.KubesphereApiTokenAuthenticator#authenticate: API token matched for user liding2 but the impersonation failed

      org.acegisecurity.userdetails.UsernameNotFoundException: User liding2 not found in directory.

      at org.acegisecurity.ldap.search.FilterBasedLdapUserSearch.searchForUser(FilterBasedLdapUserSearch.java:126)

      at hudson.security.LDAPSecurityRealm$LDAPUserDetailsService.loadUserByUsername(LDAPSecurityRealm.java:1320)

      at hudson.security.LDAPSecurityRealm$LDAPUserDetailsService.loadUserByUsername(LDAPSecurityRealm.java:1273)

      at org.acegisecurity.userdetails.UserDetailsService.lambda$toSpring$1(UserDetailsService.java:52)

      Caused: org.springframework.security.core.userdetails.UsernameNotFoundException

      at org.acegisecurity.userdetails.UsernameNotFoundException.toSpring(UsernameNotFoundException.java:51)

      at org.acegisecurity.userdetails.UsernameNotFoundException.toSpring(UsernameNotFoundException.java:34)

      at org.acegisecurity.userdetails.UserDetailsService.lambda$toSpring$1(UserDetailsService.java:54)

      at jenkins.security.ImpersonatingUserDetailsService2.loadUserByUsername(ImpersonatingUserDetailsService2.java:29)

      at hudson.model.User.getUserDetailsForImpersonation2(User.java:406)

      at hudson.model.User.getUserDetailsForImpersonation(User.java:429)

      Caused: org.acegisecurity.userdetails.UsernameNotFoundException: org.springframework.security.core.userdetails.UsernameNotFoundException: org.acegisecurity.userdetails.UsernameNotFoundException: User liding2 not found in directory.; nested exception is org.springframework.security.core.userdetails.UsernameNotFoundException: org.acegisecurity.userdetails.UsernameNotFoundException: User liding2 not found in directory.

      2，按照你前面提的三个步骤操作了，流水线又成功运行了，一会儿后又发现流水线"未运行"
      3，检查Jenkins的role-strategy/assign-roles


      4，查看devops-controller日志
      E0821 03:07:25.426300 1 pipelinerun_controller.go:205] pipelinerun-controller “msg”=“unable to run pipeline” “error”=“unexpected status code: 500” “Pipeline”=“elf” “PipelineRun”={“Namespace”:“tools6rwrl”,“Name”:“elf-gqmc7”} “namespace”=“tools6rwrl” “pipeline”=“elf”

      E0821 03:07:25.426407 1 controller.go:326] “msg”=“Reconciler error” “error”=“unexpected status code: 500” “controller”=“pipelinerun” “controllerGroup”=“devops.kubesphere.io” “controllerKind”=“PipelineRun” “name”=“elf-gqmc7” “namespace”=“tools6rwrl” “pipelineRun”={“name”:“elf-gqmc7”,“namespace”:“tools6rwrl”} “reconcileID”=“6d87a5d2-3543-40d7-94d3-ec3d14fd0bef”

      <nil>

      <nil>

      E0821 03:09:32.357775 1 devopscredential_controller.go:294] unexpected error type: not found resources, should be *restful.ServiceError

      E0821 03:09:32.357826 1 devopscredential_controller.go:167] error syncing ‘cssgzndq/gitlab-token’: failed to remove devops credential finalizer due to bad communication with Jenkins, requeuingcould not reconcile devopsProject

      E0821 03:09:32.357834 1 devopscredential_controller.go:168] error syncing ‘cssgzndq/gitlab-token’: failed to remove devops credential finalizer due to bad communication with Jenkins, requeuing

      E0821 03:09:32.428121 1 devopscredential_controller.go:294] unexpected error type: not found resources, should be *restful.ServiceError

      E0821 03:09:32.428175 1 devopscredential_controller.go:167] error syncing ‘cssgzndq/k8s-token’: failed to remove devops credential finalizer due to bad communication with Jenkins, requeuingcould not reconcile devopsProject

      E0821 03:09:32.428183 1 devopscredential_controller.go:168] error syncing ‘cssgzndq/k8s-token’: failed to remove devops credential finalizer due to bad communication with Jenkins, requeuing

      E0821 03:09:32.449208 1 devopscredential_controller.go:294] unexpected error type: not found resources, should be *restful.ServiceError

      E0821 03:09:32.449252 1 devopscredential_controller.go:167] error syncing ‘uosjmdbk/gitlab-token’: failed to remove devops credential finalizer due to bad communication with Jenkins, requeuingcould not reconcile devopsProject

      E0821 03:09:32.449260 1 devopscredential_controller.go:168] error syncing ‘uosjmdbk/gitlab-token’: failed to remove devops credential finalizer due to bad communication with Jenkins, requeuing

      E0821 03:09:32.461410 1 devopscredential_controller.go:294] unexpected error type: not found resources, should be *restful.ServiceError

      E0821 03:09:32.461456 1 devopscredential_controller.go:167] error syncing ‘uosjmdbk/k8s-token’: failed to remove devops credential finalizer due to bad communication with Jenkins, requeuingcould not reconcile devopsProject

      E0821 03:09:32.461465 1 devopscredential_controller.go:168] error syncing ‘uosjmdbk/k8s-token’: failed to remove devops credential finalizer due to bad communication with Jenkins, requeuing

      <nil>

      <nil>

      <nil>

      <nil>

      E0821 03:14:01.416187 1 pipelinerun_controller.go:205] pipelinerun-controller “msg”=“unable to run pipeline” “error”=“unexpected status code: 500” “Pipeline”=“css” “PipelineRun”={“Namespace”:“tools6rwrl”,“Name”:“css-vc4d8”} “namespace”=“tools6rwrl” “pipeline”=“css”

      E0821 03:14:01.416736 1 controller.go:326] “msg”=“Reconciler error” “error”=“unexpected status code: 500” “controller”=“pipelinerun” “controllerGroup”=“devops.kubesphere.io” “controllerKind”=“PipelineRun” “name”=“css-vc4d8” “namespace”=“tools6rwrl” “pipelineRun”={“name”:“css-vc4d8”,“namespace”:“tools6rwrl”} “reconcileID”=“7917f670-219b-4738-b3fa-311dedcd39fe”

      E0821 03:14:25.682510 1 pipelinerun_controller.go:205] pipelinerun-controller “msg”=“unable to run pipeline” “error”=“unexpected status code: 500” “Pipeline”=“elf” “PipelineRun”={“Namespace”:“tools6rwrl”,“Name”:“elf-rhgbf”} “namespace”=“tools6rwrl” “pipeline”=“elf”

      E0821 03:14:25.682684 1 controller.go:326] “msg”=“Reconciler error” “error”=“unexpected status code: 500” “controller”=“pipelinerun” “controllerGroup”=“devops.kubesphere.io” “controllerKind”=“PipelineRun” “name”=“elf-rhgbf” “namespace”=“tools6rwrl” “pipelineRun”={“name”:“elf-rhgbf”,“namespace”:“tools6rwrl”} “reconcileID”=“1489e14b-bb48-4ca7-8472-2a6ca2ee2ad6”

      5，对比devops-config，kubesphere-config，kubesphere-secret

      6，疑惑的是kubesphere-devops-system/secrets/devops-jenkins中定义的jenkins-admin-password无法登录Jenkins，重启了Jenkins之后发现也登陆不了，最终用的是默认密码登录。

          BombLing 感谢反馈，请问环境是 Kubesphere 哪个版本 ？然后麻烦对比 Kubesphere-config 和 devops-config 里的 jwtSecret 是不是一样的；

              yudong

              1，对比Kubesphere-config 和 devops-config 里的 jwtSecret 是一样的；

              2，Kubesphere是3.4.0，从3.2.1升级的；
              3，回顾时间线：应该devops项目中删除了流水线，或者删除了devops项目，导致拥有该项目的权限的用户出现“未运行”

              4，刚刚发现有devops项目未能成功删除，原有是删除secrets时卡住，使用patch才能删除；

                  BombLing 哦 如果 jwtSecret 和 password（截图中的）是一样的，并且重启过 ks-controller-manager 服务，那就没有问题了。
                  对删除 devops 项目或者流水线卡住问题，可以在 k8s 节点终端上手动操作删除；

                      yudong

                      对比了kubesphere-secret中的secret，kubesphere-config中的jwtSecret，devops-config中的 jwtSecret三者一致，也重启了ks-controller-manager，但是Jenkins的用户显示

                      在什么情况下会触发更新Jenkins的role-strategy/assign-roles？

                          BombLing 哦，同步用户信息在修改用户密码时会触发，您可以试着修改下用户密码。

                              @yudong
                              ks: 3.3.2
                              k8s: 1.24.9
                              现象是在多分支流水线中，经常会出现个别流水线一直未运行的情况，其它流水线都正常，重启相关组件都未解决。通过删除重新创建就可以重新运行。
                              还有一个现象是流水线已经执行完成，jenkins中也已经结束了，但是ks上流水线的状态一直是运行中

                                littlejiancc 可能是同时运行的流水线太对了，这些流水线在排队等待中；可以看下 kubesphere-devops-worker 空间下是否有多个 pod ?

                                    13 天 后
                                    8 天 后

                                    yudong 您好，我是在jenkins安装插件后，流水线点击构建提示未运行，您说的这3步我不清楚怎么操作

                                        runaway

                                        kubectl get pod -n kubesphere-devops-system

                                        看下所有 pod 都启动了吗，包括 devops-controller 和 jenkins

                                        如果是，可以看下 devops-controller 中是否有什么报错。

                                            chilianyi 感谢回复，我说明一下我这边的情况，k8s是腾讯云的TKE，在TKE上安装kubesphere，因为需要用到钉钉构建通知，我就找到jenkins的web界面，进去安装了一个dingtalk插件，安装后会重启jenkins，重启之后，所有项目就不能构建了。

                                            1. 使用 devops-config 中这个 devops/password 替换 secret kubesphere-secret 中的 token；

                                            2. 使用 devops-config 中这个 devops/password，替换 kubesphere-config 中的 devops/password；

                                            3. 重启 Deployment ks-controller-manager ；

                                            以上三步操作我后来也找到了但是执行后，还是点击构建流水线，流水线未运行

                                              chilianyi pod 启动了，包括 devops-controller 和 jenkins

                                              devops-controller日志中的提示：

                                                  chilianyi

                                                  操作系统信息
                                                  腾讯云：TencentOS Server 3.1(TK4) 8C/16G 3台

                                                  （腾讯云容器服务，创建集群时创建的）

                                                  Kubernetes版本信息
                                                  版本：1.22.5

                                                  容器运行时
                                                  containerd

                                                  KubeSphere版本信息
                                                  v3.3.2，通过 ks-installer 执行最小化部署