When searching for resources in a cluster that has stopped synchronizing,
alert the user that the resources getted may be inaccurate

What would you like to be added?

I the License file, the copyright is:
Copyright [yyyy] [name of copyright owner]

suggest to change the license copyright to correct one.

Why is this needed?

perfesional

1. Clusterpedia uses LabelSelector to filter data, but you need to manually build the Options object
2. We need to use a more native way to call, like client-Go
3. You only need to provide akubeconfig to get the clientset of client-go

Here's a project to solve the above problem.

Welcome to discuss it here👏

Thanks, clusterpedia is such a wonderful job, but I think it's something helpful like CONTRIBUTING.md to encourage developers to contribute to this project. I can provide some help.

When using fieldselector, the log will report an error： "Couldn't parse request" err="invalid selector: 'involvedObject.uid in xxx"
The detail is below:

What would you like to be added?

1.将被接入集群的kuconfig文件转换为base64格式
2.使用转换的kuconfig接入集群
预期结果：接入成功
实际结果：接入失败

Why is this needed?

被接入集群的kubeconfig的userclient相关配置项需转换为base64

It is difficult to search for resources by their owner relationships in kubernetes.
clusterpedia will first introduce this experimental feature in internalstorage.

There are two types of owner functions.

• Query based on the owner's UID
• Query by owner's name, and support for setting the owner's GVR

In addition, owner filtering supports the ability to raise the seniority to meet the function of ancestor query.

Based on the above features, we plan to add four labels to the internalstorage first:

• internalstorage.clusterpedia.io/owner-uid
• internalstorage.clusterpedia.io/owner-name
• internalstorage.clusterpedia.io/owner-gvr
• internalstorage.clusterpedia.io/owner-seniority

Experimental function, the url query is not available for now

internalstorage search and sql examples

search pods base on the UID of the owner, label selector is
internalstorage.clusterpedia.io/owner-uid=018442f8-8e14-11eb-9f9e-0242ac130002

SELECT * FROM `resources` WHERE `group` = "" AND `version` = "v1" AND `resource` = "pods" AND `cluster` = "cluster-dce3" AND \
    owner_uid  = "018442f8-8e14-11eb-9f9e-0242ac130002"

search pods based on the UID of the grandfather owner, label selector is
internalstorage.clusterpedia.io/owner-uid=018442f8-8e14-11eb-9f9e-0242ac130002,internalstorage.clusterpedia.io/owner-seniority=1

SELECT * FROM `resources` WHERE `group` = "" AND `version` = "v1" AND `resource` = "pods" AND `cluster` = "cluster-dce3" AND \
        owner_uid IN (SELECT `uid` FROM `resources` WHERE `cluster` = "cluster-dce3" AND owner_uid = "018442f8-8e14-11eb-9f9e-0242ac130002")

An owning object must be in the same namespace as the dependent, or be cluster-scoped.
search pods based on the name of the grandfather owner, label selecttor is
internalstorage.clusterpedia.io/owner-name=dmp-eureka-enhance-1,internalstorage.clusterpedia.io/owner-seniority=1

SELECT * FROM `resources` WHERE `group` = "" AND `version` = "v1" AND `resource` = "pods" AND `cluster` = "cluster-dce3" AND \
    owner_uid IN (SELECT `uid` FROM `resources` WHERE `cluster` = "cluster-dce3" AND \
        owner_uid IN (SELECT `uid` FROM `resources` WHERE `cluster` = "cluster-dce3" AND `name` = "dmp-eureka-enhance-1" AND `namespace` = "dmp"))

The cluster synchro of pediacluster may fail to be. created due to many reasons, such as network problems, or authentication configuration....

Clusterpedia need to update the specific information to pediacluster's Conditions
eg.

status:
  conditions:
  - type: ClusterSynchro
     status: "False"
     reason: "InitialFailed"
     message: "some error message"
  - message: ""
    reason: Pending
    status: "False"
    type: Ready

When clustersynchro manager is shutdown for some reason, the status of the pediaclusters in the cluster need to be updated to indicate that the cluster synchro has been stopped

What would you like to be added?

.kube/config文件中client证书配置为路径时，手动替换为base64形式，在接入clusterpedia时，提示“file name to long”

Why is this needed?

对配置读取进行优化

Currently, if need to return the RemainingItemCount, we need to set withRemaingCount in the Search Label or url query #33

It is now possible to allow all requests to return the RemainingItemCount by reusing the feature gate of the kube apiserver -- RemainingItemCount

What would you like to be added?

Support fuzzy query through the name of the resource and the range of creation time.

Why is this needed?

This is commonly used in daily use. And it is convenient for our development and operation and maintenance personnel

What would you like to be added?

PdiaCluster can directly reference fields from existing resources for cluster validation.

eg.

spec:
  apiserver: https://10.6.100.100:6443
  caRef:
    apiVersion: v1
    kind: Configmap
    name: cluster-1
    fieldPath: ".data.ca"

Why is this needed?

Currently we need to fill caData, tokenData and other information in the PediaCluster,
but often these information may already exist in other resources,
it would be very convenient if we can get the authentication information by referring to the fields of these resources.

What would you like to be added?

Support find out which Deploy/StatefulSet/DaemonSet a pod belongs to.

Why is this needed?

We would like to have a pod list related to a specific service, and the workload(deploy/statefulset/daemonset) related to the pod. Of course we can get the related workloads with the service selector, but we can only get pods and workloads separately, we can't tell directly which workload one pod belongs to.

What would you like to be added?

1.Retrieve resources and set the number of pagination to 5 and and set offset to 0
2.set offset to 1
3.set offset to 2
Actual results：
1.1-5 resources retrieved
2.2-6 resources retrieved
3.3-7 resources retrieved

Why is this needed?

Expected results：
1.1-5 resources retrieved
2.6-10 resources retrieved
3.11-15 resources retrieved

What would you like to be added?

Support wildcard (like "*") for resources to sync, in the following format:

# all resources in all groups
syncResources:
- group: "*"
  resources:
  - "*"

# all resources in group apps
syncResources:
- group: "apps"
  resources:
  - "*"

Why is this needed?

we need to list all resources in a cluster dynamically (the resource types might be changing)

support list field filtering

kubectl get po --field-selector="spec.containers[].name!=container1"

kubectl get po --field-selector="spec.containers[].name == container1"

kubectl get po --field-selector="spec.containers[1].name in (container1,container2)"

PostgreSQL

SELECT *
FROM "resources",
     jsonb_array_elements(object -> 'status' -> 'conditions') as obj
WHERE "group" = 'apps'
  AND "resource" = 'deployments'
  AND "version" = 'v1'
  AND obj ->> 'status' = 'True'
ORDER BY namespace, name
LIMIT 10;

MySQL

SELECT *
FROM resources
WHERE `group` = 'apps'
  AND resource = 'deployments'
  AND version = 'v1'
  AND json_contains(object, '{"status": "True"}', '$.status.conditions')
ORDER BY namespace, name
LIMIT 10;

What happened?

指定集群也可以使用 --cluster 来指定

文档最后面多了个"

kubectl --cluster cluster-1 get deployments"

What did you expect to happen?

修改文档，去掉"

How can we reproduce it (as minimally and precisely as possible)?

文档地址：https://clusterpedia.io/zh-cn/docs/usage/search/multi-cluster/

在指定集群的说明里面

指定集群也可以使用 --cluster 来指定
kubectl --cluster cluster-1 get deployments"

Anything else we need to know?

No response

Clusterpedia Version

# replace the pod name of clusterpedia apiserver
$ kubectl -n clusterpedia-system exec {clusterpedia-apiserver-pod-name} -- apiserver --version=raw
# paste output here

# replace the pod name of clusterpedia clustersynchro-manager
$ kubectl -n clusterpedia-system exec {clustersynchro-manager-pod-name} -- clustersynchro-manager --version=raw
# paste output here

Host Kubernetes version

$ kubectl version
# paste output here

What would you like to be added?

Should we add license header in go file

Why is this needed?

It's very common. I'm not very sure the usage.

The current filter selector supports filtering on fields and values (eg. --filter-selector="spec.replicas=1"), but in some cases we need to compare two fields within a resource.
eg. --field-selector="spec.replicas == .status.readyReplicas"

Currently, when comparing between two fields, it looks like it needs to support only == and !=.
There is also the question of how to better distinguish between field comparisons and field-value comparisons.

What would you like to be added?

In the past few releases, both featrue PruneLastAppliedConfiguration and PruneManagedFields have been stable and mature. Will we update these featuregate in V0.0.3?

Why is this needed?

feature upgrade

What would you like to be added?

Many complex search queries are currently available, such as field selector and search by owner, but sometimes users have more customizable query needs .

To make the query more powerful, the plan wants to add an exclusive query feature to internalstorage that supports users passing a piece of SQL and then splicing the user's SQL after the WHERE.

Of course, in order to be compatible with the generic Search Label, which is still available, the SQL will be spliced after the sql according to the Search Label, but before the ORDER BY.

SELECT objects WHERE group='apps' AND version='v1' AND resource='deployments' AND <user sql> ORDER BY `id`

There are two current questions that need to be resolved

• How to prevent the injection of the passed SQL string
• How to be compatible with the Kubernetes API, after all, Label Selector's value checking is rather strict

Also the different syntax between postgres and mysql may need to be handled by the user depending on the storage component they choose .
The clusterpedia does not handle the compatibility of SQL pieces in different storage components.

However, the final generated SQL may be carried in the response header to allow the user to know the SQL statement

The clustersynchro prunes fields based on feature gate when storing resources.
issue: support for pruning manageFields and last-applied-configuration fields
pr: prune managedFields and last-applied-configuration

Since the managedFields field is typically used very infrequently in resource searches, pruning prior to storage can also reduce the strain on storage and data transfer.

However, this may result in inconsistencies between the stored resources and the resource data in the member clusters.

Considering whether the clusterpedia apiserver should also support pruning to ensure that the complete data is stored and then pruned when the data is returned

During resource synchronization, the informer reflector cannot report errors externally if a resource(custom resource) is deleted, or after not having the correct resource permissions.

We need to make the reflector able to tell the resource synchro more information

What happened?

Search svc directly, service-nodeport has CLUSTER-IP

[root@dce-10-6-215-215 ~]# kubectl get svc -n zouzou
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service-clusterip ClusterIP 172.31.88.88 8080/TCP 4h44m
service-headliness ClusterIP None 9090/TCP 4h44m
service-nodeport NodePort 172.31.132.180 8888:30033/TCP 4h44m

By specifying a name search in clusterpedia, the cluster IP is not searched

[root@dce-10-6-215-215 ~]# kubectl --cluster clusterpedia get svc -n zouzou -l "search.clusterpedia.io/names=service-nodeport"
CLUSTER NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
cluster-215-215 service-nodeport NodePort 8888:30033/TCP 4h44m

What did you expect to happen?

SVC searched through clusterpedia also shows CLUSTER-IP

How can we reproduce it (as minimally and precisely as possible)?

[root@dce-10-6-215-215 ~]# kubectl get svc -n zouzou
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service-clusterip ClusterIP 172.31.88.88 8080/TCP 4h44m
service-headliness ClusterIP None 9090/TCP 4h44m
service-nodeport NodePort 172.31.132.180 8888:30033/TCP 4h44m
[root@dce-10-6-215-215 ~]# kubectl --cluster clusterpedia get svc -n zouzou -l "search.clusterpedia.io/names=service-nodeport"
CLUSTER NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
cluster-215-215 service-nodeport NodePort 8888:30033/TCP 4h44m

Anything else we need to know?

No response

Clusterpedia Version

# replace the pod name of clusterpedia apiserver
$ kubectl -n clusterpedia-system exec {clusterpedia-apiserver-pod-name} -- apiserver --version=raw
# paste output here

# replace the pod name of clusterpedia clustersynchro-manager
$ kubectl -n clusterpedia-system exec {clustersynchro-manager-pod-name} -- clustersynchro-manager --version=raw
# paste output here

Host Kubernetes version

$ kubectl version
Client Version: version.Info{Major:"1", Minor:"18", GitVersion:"v1.18.20", GitCommit:"1f3e19b7beb1cc0110255668c4238ed63dadb7ad", GitTreeState:"clean", BuildDate:"2021-06-16T12:58:51Z", GoVersion:"go1.13.15", Compiler:"gc", Platform:"linux/amd64"}
Server Version: version.Info{Major:"1", Minor:"18", GitVersion:"v1.18.20", GitCommit:"7c28a83e67cb506f9df948d7b23698f8f9c279b8", GitTreeState:"clean", BuildDate:"2021-07-15T08:58:53Z", GoVersion:"go1.13.15", Compiler:"gc", Platform:"linux/amd64"}

What would you like to be added?

In these cases, our experience is very unfriendly:

• when we need to synchronize multiple clusters at once, we need to populate a large amount of resource information, including very basic resources. e.g: pod, node, deployment
• if you want to add or delete multiple pediaCluster synchronization resources at the same time

we can use a resource template, when we modify the template, all pediaCluster in the cluster will change dynamically. Exemptions can be used for pediaCluster that needs to be excluded.

Why is this needed?

We can more easily modify all pediaCluster resources in the cluster.

What happened?

When we install clusterpedia, it restarts for more than 5 times

What did you expect to happen?

The Clusterpedia not restart, it wait for mysql ready.

How can we reproduce it (as minimally and precisely as possible)?

Install Clusterpeida using Kpanda

Anything else we need to know?

We can use initContainer to check mysql ready. Eg:

• https://stackoverflow.com/questions/65646499/how-to-use-an-init-container-to-check-if-mysql-is-ready-for-connections
• https://www.magalix.com/blog/kubernetes-patterns-the-init-container-pattern

Clusterpedia Version

# replace the pod name of clusterpedia apiserver
$ kubectl -n clusterpedia-system exec {clusterpedia-apiserver-pod-name} -- apiserver --version=raw
# paste output here

# replace the pod name of clusterpedia clustersynchro-manager
$ kubectl -n clusterpedia-system exec {clustersynchro-manager-pod-name} -- clustersynchro-manager --version=raw
# paste output here

Host Kubernetes version

$ kubectl version
# paste output here

What would you like to be added?

Why is this needed?

As described above.

• deploy api server
• deploy clustersynchro manager
• deploy a usable database, mysql or postgresql
• relevant RBAC and serviceAccout

Storage Component: mysql

kubectl --cluster clusterpedia get deploy --field-selector="status.availableReplicas==1" -A

The sql generated by the sql builder is

SELECT * FROM `resources` WHERE `group` = 'apps' AND `resource` = 'deployments' AND `version` = 'v1' AND cluster = 'test-cluster-2' AND namespace = 'default' AND \
     JSON_EXTRACT(`object`,'$."status"."availableReplicas"') == '1' LIMIT 500

In this case, only if the field value is the string '1'/"1" will be searched for.
For example, this will work:

kubectl --cluster pedia get deploy --field-selector="metadata.annotations['deployment.kubernetes.io/revision']=1" 

resources:

apiVersion: apps/v1
kind: Deployment
metadata:
  annotations:
    deployment.kubernetes.io/revision: "1"

What would you like to be added?

If we deploy Clusterpedia using kubectl apply instead of helm, having to find the CRD files from the /charts/_crd/ directory would be a hassle.

We can soft-link the files to the /deploy directory.

Why is this needed?

Be more friendly to deploy Clusterpedia.

What happened?

查询 ns 的时候，限制了条数为 2，进行查询的时候，报错

[root@dce-10-6-215-215 pod-dir]# kubectl --cluster cluster-example get namespace --chunk-size 2
error: the server doesn't have a resource type "namespace"

What did you expect to happen?

可以查询出 ns，并且返回两条数据

How can we reproduce it (as minimally and precisely as possible)?

kubectl --cluster cluster-example get namespace --chunk-size 2

Anything else we need to know?

No response

Clusterpedia Version

# replace the pod name of clusterpedia apiserver
$ kubectl -n clusterpedia-system exec {clusterpedia-apiserver-pod-name} -- apiserver --version=raw
# paste output here

# replace the pod name of clusterpedia clustersynchro-manager
$ kubectl -n clusterpedia-system exec {clustersynchro-manager-pod-name} -- clustersynchro-manager --version=raw
# paste output here

Host Kubernetes version

$ kubectl version
Client Version: version.Info{Major:"1", Minor:"18", GitVersion:"v1.18.20", GitCommit:"1f3e19b7beb1cc0110255668c4238ed63dadb7ad", GitTreeState:"clean", BuildDate:"2021-06-16T12:58:51Z", GoVersion:"go1.13.15", Compiler:"gc", Platform:"linux/amd64"}
Server Version: version.Info{Major:"1", Minor:"18", GitVersion:"v1.18.20", GitCommit:"7c28a83e67cb506f9df948d7b23698f8f9c279b8", GitTreeState:"clean", BuildDate:"2021-07-15T08:58:53Z", GoVersion:"go1.13.15", Compiler:"gc", Platform:"linux/amd64"}

Users are likely to use apis, such as search labels, and clusterpedia-io/client-go will also need to use search labels.

The import of kubernetes into clusterpedia causes complex package dependencies that make it difficult for users to directly import clusterpedias/pkg/apis to use

In process of deplyment the clustersynchro manager, The following error message appears:

Error: unknown flag: --feature-gates PruneManagedFields

What would you like to be added?

Use kubeconfig directly, instead of filling in ca, cert, key, etc. separately in PediaCluster

Why is this needed?

Using kubeconfig will make the creation of pediacluster much easier.

What would you like to be added?

The Features List can not directly make people to adopt it.

To a man with a hammer, everything looks like a nail.

Now we have the hammer, but how to promote to the world more effectively , so those nails can find this hammer?

1. SOE maybe one way. Maybe you can describe as many as possible scenario ,which this hammer can help .
2. Treat the project as a product, do everything those product manager will do : analysis and advocation ..etc

Just my two cents.
The only "nail" I can figure out is the tools acting as cluster-gateway like PaaS. But I believe this hammer could fit more nail.

Why is this needed?

as above.

What happened?

What did you expect to happen?

When sorting the resource_version field in the cluster, the query results are correct

How can we reproduce it (as minimally and precisely as possible)?

kubectl --cluster clusterpedia get pods -l "search.clusterpedia.io/orderby in (resource_version_desc)"
kubectl --cluster clusterpedia get pods -l "search.clusterpedia.io/orderby in (resource_version)"

Anything else we need to know?

No response

Clusterpedia Version

# replace the pod name of clusterpedia apiserver
$ kubectl -n clusterpedia-system exec {clusterpedia-apiserver-pod-name} -- apiserver --version=raw
# paste output here

# replace the pod name of clusterpedia clustersynchro-manager
$ kubectl -n clusterpedia-system exec {clustersynchro-manager-pod-name} -- clustersynchro-manager --version=raw
# paste output here

Host Kubernetes version

$ kubectl version
Client Version: version.Info{Major:"1", Minor:"18", GitVersion:"v1.18.20", GitCommit:"1f3e19b7beb1cc0110255668c4238ed63dadb7ad", GitTreeState:"clean", BuildDate:"2021-06-16T12:58:51Z", GoVersion:"go1.13.15", Compiler:"gc", Platform:"linux/amd64"}
Server Version: version.Info{Major:"1", Minor:"18", GitVersion:"v1.18.20", GitCommit:"7c28a83e67cb506f9df948d7b23698f8f9c279b8", GitTreeState:"clean", BuildDate:"2021-07-15T08:58:53Z", GoVersion:"go1.13.15", Compiler:"gc", Platform:"linux/amd64"}

Server-Side Apply stores field management in the manageFields, which may not be useful during resource search.
This field needs to be pruned according to the user's needs.

The kubectl.kubernetes.io/last-applied-configuration annotation also requires a feature gate to be provide to prune.

kind: Deployment
metadata:
  annotations:
    deployment.kubernetes.io/revision: "5"
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"apps/v1","kind":"Deployment","metadata":{"annotations":{},"labels":{"app":"cluster-autoscaler"},"name":"poc-autoscaler","namespace":"default"},"spec":{"replicas":1,"selector":{"matchLabels":{"app":"cluster-autoscaler"}},"template":{"metadata":{"labels":{"app":"cluster-autoscaler"}},"spec":{"containers":[{"args":["--cloud-provider=clusterapi","--kubeconfig=/etc/kubeconfig/value","--clusterapi-cloud-config-authoritative","--node-group-auto-discovery=clusterapi:clusterName=poc"],"command":["/cluster-autoscaler"],"image":"daocloud.io/daocloud/cluster-autoscaler:v1.20.1","name":"cluster-autoscaler","volumeMounts":[{"mountPath":"/etc/kubeconfig","name":"kubeconfig"}]}],"serviceAccountName":"cluster-autoscaler-management","terminationGracePeriodSeconds":10,"tolerations":[{"effect":"NoSchedule","key":"node-role.kubernetes.io/master"}],"volumes":[{"name":"kubeconfig","secret":{"secretName":"poc-kubeconfig"}}]}}}}

When updating the cluster status, DeepEqual will be called first to compare the cluster status to see if it is the same, and if it is, the update will be skipped.

However, there are now three places where the cluster status will always be unequal:

1. cluster.Status is not ptr, but status is ptr
2. ClusterGroupStatus.Resources is not sorted.

   clusterpedia/pkg/synchromanager/clustersynchro/cluster_synchro.go

   Line 410 in 9fb3c25

   groupStatus.Resources = append(groupStatus.Resources, *resourceStatus)

3. metav1.Now() is too precise, and should use the metav1.Now().Rfc3339Copy()

What happened?

If the release name contains the name of a storage component, then clusterpedia will not start successfully.

realease name: clusterpedia-with-postgresql

$ helm install clusterpedia-with-postgresql  -n clusterpedia-system --create-namespace  --set persistenceMatchNode=caiwei-master --set installCRDs=true .

╭─icebergu@IceberGudeMBP ~/clusterpedia/charts  ‹add-helm*›
╰─➤  kubectl -n clusterpedia-system get pods
NAME                                                              READY   STATUS             RESTARTS      AGE
clusterpedia-with-postgresql-0                                    1/1     Running            0             13s
clusterpedia-with-postgresql-apiserver-b96d4c599-4tlcs            0/1     CrashLoopBackOff   1 (11s ago)   13s
clusterpedia-with-postgresql-clustersynchro-manager-7f76cd2sgch   0/1     CrashLoopBackOff   1 (11s ago)   13s

╭─icebergu@IceberGudeMBP ~/clusterpedia/charts  ‹add-helm*›
╰─➤  kubectl -n clusterpedia-system logs clusterpedia-with-postgresql-apiserver-b96d4c599-4tlcs  
Failed to init storage: failed to connect to `host=clusterpedia-with-postgresql-postgresql user=postgres database=clusterpedia`: hostname resolving error (lookup clusterpedia-with-postgresql-postgresql on 10.96.0.10:53: no such host)

Found that the svc name provided by postgres does not match the host in internalstorage config:

╭─icebergu@IceberGudeMBP ~/clusterpedia/charts  ‹add-helm*›
╰─➤  kubectl -n clusterpedia-system get svc
NAME                                     TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)    AGE
clusterpedia-with-postgresql             ClusterIP   10.108.157.173   <none>        5432/TCP   4m29s
clusterpedia-with-postgresql-apiserver   ClusterIP   10.99.111.2      <none>        443/TCP    4m29s
clusterpedia-with-postgresql-hl          ClusterIP   None             <none>        5432/TCP   4m29s

╭─icebergu@IceberGudeMBP ~/clusterpedia/charts  ‹add-helm*›
╰─➤  kubectl -n clusterpedia-system get cm clusterpedia-with-postgresql-internalstorage -o yaml
apiVersion: v1
data:
  internalstorage-config.yaml: |-
    type: "postgres"
    host: "clusterpedia-with-postgresql-postgresql"
    port: 5432
    user: postgres
    database: clusterpedia
kind: ConfigMap

The host of the internstorage configmap has an additional postgres suffix.
The problem should be that the host of the internalstorage configmap generated in charts is not the same as the svc name of postgresql.

What did you expect to happen?

The host of internalstorage is the same as the svc name of postgresql, and the clusterpedia will start successfully.

How can we reproduce it (as minimally and precisely as possible)?

output template:

$ helm template clusterpedia-with-postgresql --set persistenceMatchNode=None  .
# Source: clusterpedia/charts/postgresql/templates/primary/svc.yaml
apiVersion: v1
kind: Service
metadata:
  name: clusterpedia-with-postgresql
  namespace: default
spec:
  type: ClusterIP
  ports:
    - name: tcp-postgresql
      port: 5432
      targetPort: tcp-postgresql
      nodePort: null
  selector:
    app.kubernetes.io/name: postgresql
    app.kubernetes.io/instance: clusterpedia-with-postgresql
    app.kubernetes.io/component: primary
# Source: clusterpedia/templates/internalstorage-configmap.yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: clusterpedia-with-postgresql-internalstorage
  namespace: default
data:
  internalstorage-config.yaml: |-
    type: "postgres"
    host: "clusterpedia-with-postgresql-postgresql"
    port: 5432
    user: postgres
    database: clusterpedia

Anything else we need to know?

No response

Clusterpedia Version

# replace the pod name of clusterpedia apiserver
$ kubectl -n clusterpedia-system exec {clusterpedia-apiserver-pod-name} -- apiserver --version=raw
# paste output here

# replace the pod name of clusterpedia clustersynchro-manager
$ kubectl -n clusterpedia-system exec {clustersynchro-manager-pod-name} -- clustersynchro-manager --version=raw
# paste output here

Host Kubernetes version

$ kubectl version
# paste output here

What would you like to be added?

Support synchronization of all CRD resources by a field designed in pediacluster.

Why is this needed?

It is useful for front-end display and viewing of all CRD resources.

For the synchronization of custom resources is developed in two stages：

• The default collection policy, where the user specifies the version to be synchronized,If no version is specified, then the top three versions of the priority order in the member cluster will be synchronized by default. #10
• When crds change, the resources that need to be synchronized are renegotiated. #136
• The clustersynchro manager reconciles the sync version and storage version that should be synchronized by the member cluster

What would you like to be added?

Allows resources to be synchronized using the singular/plural of the resource and can be case-insensitive

Why is this needed?

In kubernetes both singular and plural are available to query the data, which increases fault tolerance

What would you like to be added?

1.使用获取被接入集群ca和token方式接入集群
2.获取ca和token失败

Why is this needed?

将被接入集群的kubeconfig中的namespace替换为default能正常获取

参考

What would you like to be added?

1.使用A.yaml文件接入集群A
2.复制A.yaml文件，修改文件名和集群名为B
3.使用B.yaml文件接入集群
预期结果：提示集群已存在
实际结果：接入集群B

Why is this needed?

对接入的yaml文件进行判断，若存在相同的集群接入，则接入失败，提示该集群已接入到clusterpedia

What happened?

[root@dce-10-6-215-215 ~]# kubectl --cluster clusterpedia get pod
CLUSTER NAME READY STATUS RESTARTS AGE
cluster-178-220 hello-79dbf8f975-fhkcl 1/1 Running 0 128m
cluster-example cm-env-test-pod 0/1 Completed 0 5d1h
cluster-example ng1-849588dbb9-dqc7r 1/1 Running 3 (19h ago) 2d16h
cluster-178-220 nginx-f89759699-rkgkg 1/1 Running 0 130m
cluster-128-168 dao-2048-4-9cdc87d65-7hbdb 1/1 Running 7 (36d ago) 210d
cluster-128-168 dao-2048-6d6dfbb6ff-cjmps 1/1 Running 6 (36d ago) 210d
cluster-128-168 my222-dao-2048-d96b45f57-hpcgn 1/1 Running 8 (36d ago) 210d

指定了某个具体的集群名称提示找不到

[root@dce-10-6-215-215 ~]# kubectl --cluster cluster-128-168 get pod
error: no server found for cluster "cluster-128-168"
[root@dce-10-6-215-215 ~]#

What did you expect to happen?

可以根据某个集群名称找到对应的 pod

How can we reproduce it (as minimally and precisely as possible)?

指定了某个具体的集群名称提示找不到

[root@dce-10-6-215-215 ~]# kubectl --cluster cluster-128-168 get pod
error: no server found for cluster "cluster-128-168"

Anything else we need to know?

No response

Clusterpedia Version

# replace the pod name of clusterpedia apiserver
$ kubectl -n clusterpedia-system exec {clusterpedia-apiserver-pod-name} -- apiserver --version=raw
# paste output here

# replace the pod name of clusterpedia clustersynchro-manager
$ kubectl -n clusterpedia-system exec {clustersynchro-manager-pod-name} -- clustersynchro-manager --version=raw
# paste output here

Host Kubernetes version

$ kubectl version
Client Version: version.Info{Major:"1", Minor:"18", GitVersion:"v1.18.20", GitCommit:"1f3e19b7beb1cc0110255668c4238ed63dadb7ad", GitTreeState:"clean", BuildDate:"2021-06-16T12:58:51Z", GoVersion:"go1.13.15", Compiler:"gc", Platform:"linux/amd64"}
Server Version: version.Info{Major:"1", Minor:"18", GitVersion:"v1.18.20", GitCommit:"7c28a83e67cb506f9df948d7b23698f8f9c279b8", GitTreeState:"clean", BuildDate:"2021-07-15T08:58:53Z", GoVersion:"go1.13.15", Compiler:"gc", Platform:"linux/amd64"}

What would you like to be added?

Support fuzzy query through the name of the resource

Why is this needed?

This is commonly used in daily use. And it is convenient for our devops to search resource!

Currently, then resources(resources,collectionresources) under pedia.clusterpedia.io are used for searching multi-cluster resources,
but the prefix pedia does not make any sense and pedia.clusterpedia.io provides the main functionality of clusterpedia,
so it is commended to change pedia.clusterpedia.io to clusterpedia.io.

In this case, the search url will be modified:

kubectl get --raw="/apis/pedia.clusterpedia.io/v1alpha1/resources/apis/apps/v1/deployments"

to

kubectl get --raw="/apis/clusterpedia.io/v1alpha1/resources/apis/apps/v1/deployments"

After the modification, the current clusterpedia-provided groups include:

• clusterpedia.io
• clusters.clusterpedia.io

Support setting the resource version in the reflector before resource synchronizatiion.

In the weak network environment, the start and stop of resource synchronization may be more frequent,
so setting the last resource version can avoid the full list