RHOCP 4.14 或更低版本Pods在namespace openshift-marketplace下频繁的created killed-CSDN博客

本文链接：https://blog.csdn.net/qq_32789063/article/details/145993471

环境

Red Hat OpenShift Container Platform (RHOCP)
4.10
4.11
4.12
4.13
4.14

Issue

OpenShift Container Platform 4.14 至 4.10 中，位于 openshift-marketplace 命名空间中的 Pod 经常被创建和终止。
pod 的 openshift-marketplace 命名空间中的多个事件被创建和终止。

31m         Normal   AddedInterface   pod/redhat-operators-name      Add eth0 [172.0.0.0/23] from ovn-kubernetes
31m         Normal   Pulling          pod/redhat-operators-name      Pulling image "registry.redhat.io/redhat/redhat-operator-index:v4.12"
31m         Normal   Pulled           pod/redhat-operators-name      Successfully pulled image "registry.redhat.io/redhat/redhat-operator-index:v4.12" in 1.562343266s (1.562350584s including waiting)
31m         Normal   Created          pod/redhat-operators-name      Created container registry-server
31m         Normal   Started          pod/redhat-operators-name      Started container registry-server
31m         Normal   Killing          pod/redhat-operators-name      Stopping container registry-server

解决

警告：未经 Red Hat OpenShift 技术支持指导，请勿执行这些步骤。

注意：禁用默认 CatalogSource 意味着它不再由 Marketplace Operator 管理，而是由集群管理员负责管理。此外，如果您已经通过默认目录安装了操作员，那么将会出现如下事件，还安装了operator 抱怨Catlogsource被删除。

constraints not satisfiable: no operators found from catalog certified-operators in namespace openshift-marketplace referenced by subscription name-operator, subscription name-operator exists

备份命名空间 openshift-marketplace 中的默认 Catlogsource

for i in `oc get catalogsources --no-headers -n openshift-marketplace | awk '{print $1}'` ; do oc get catalogsources -n openshift-marketplace $i -o yaml > $i.yaml; done

验证文件和内容：

certified-operators.yaml
community-operators.yaml
redhat-marketplace.yaml
redhat-operators.yaml

禁用默认的 catalogsource

 oc patch OperatorHub cluster --type json -p '[{"op": "add", "path": "/spec/disableAllDefaultSources", "value": true}]' 

oc patch operatorhub/cluster --type=merge -p '{"spec":{"sources":[{"disabled":true,"name":"redhat-operators"}]}}' 
Verify no default catalogsource should exist.

验证不应存在默认catalogsource。

oc get catalogsource -n openshift-marketplace

编辑我们在第一步中备份的文件并更改相应catalogsource的名称，例如：

vi redhat-operators.yaml

name: redhat-operators     << change this name as like redhat-operators-1 . Ultimate goal is to modify each catalogsource name from default to new name.

同时删除 Generation/uid 行
现在对所有文件执行相同的操作。

更改默认名称后，让我们使用 oc create -f 创建自定义catalog

oc create -f certified-operators.yaml
oc create -f community-operators.yaml
oc create -f redhat-marketplace.yaml
oc create -f redhat-operators.yaml

使用以下命令验证必须创建新的 Catalogsource 和相应的 Pod。

# oc get catsrc -n openshift-marketplace;oc get pods -n openshift-marketplace
NAME                    DISPLAY               TYPE   PUBLISHER   AGE
certified-operators-1   Certified Operators   grpc   Red Hat     86s
community-operators-1   Community Operators   grpc   Red Hat     5m5s
redhat-marketplace-1    Red Hat Marketplace   grpc   Red Hat     5m5s
redhat-operators-1      Red Hat Operators     grpc   Red Hat     5m5s
NAME                                    READY   STATUS    RESTARTS   AGE
certified-operators-1-wfz9g             1/1     Running   0          86s
community-operators-1-8f2js             1/1     Running   0          5m5s
marketplace-operator-6f9748c96f-b75dz   1/1     Running   1          3d20h
redhat-marketplace-1-mjzzr              1/1     Running   0          5m4s
redhat-operators-1-9cbnr                1/1     Running   0          5m4s

现在继续更改 updateStrategy.registryPoll.interval 的值：每个 catalosource 为 10m0s（例如 87h）。

# oc edit CatalogSource <CatalogSource_name>
spec:
  updateStrategy:
    registryPoll:
      interval: 87h

根本原因：

此问题是由 OpenShift Container Platform 4.13-4.10 中的一个已知错误引起的，更多信息可在此处找到：KCS 7016065。