Recreate clusters after helm delete operator

[neelasha-09]

Hi Team,

While trying to delete the resources using helm delete --purge for postgres-operator and recreating the cluster with the same name we could see the below error.

reason: 'could not create pod disruption budget: poddisruptionbudgets.policy \"postgres-postgres-operator-cluster-pdb\" already exists"  

Could you please have a look ?

[FxKu]

Deleting the operator does not delete running Postgres clusters. That's actually a feature 😃

[neelasha-09]

Hi Felix Deleting the operator and recreating the *operator* with the same name produces the error.

…

On Mon, Jul 13, 2020 at 6:19 PM Felix Kunde ***@***.***> wrote: Deleting the operator does not delete running Postgres clusters. That's actually a feature 😃 — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#1057 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AP5SFQQYZT4IQMOCSBJFGITR3L7FDANCNFSM4OYNGDPQ> .

[FxKu]

If you run helm delete postgres-operator you delete the controller for Postgres CRDs but not the postgres clusters. If you check with kubectl get postgresqls you should still be able to see your postgres-postgres-operator-cluster listed in Running state. Therefore, trying to create that cluster again will fail with the message you've posted. It can happen that a DELETE event does not clean up all resources, see #498. I don't know if this happens in your case.

To properly remove a Postgres cluster run kubectl delete postgresql postgres-postgres-operator-cluster. Then all elements incl. PDB, endpoints etc. should be removed.

[neelasha-09]

Thanks Felix :)

…

On Tue, Jul 14, 2020 at 8:05 PM Felix Kunde ***@***.***> wrote: If you run helm delete postgres-operator you delete the controller for Postgres CRDs but not the postgres clusters. If you check with kubectl get postgresqls you should still be able to see your postgres-postgres-operator-cluster listed in Running state. Therefore, trying to create that cluster again will fail with the message you've posted. It can happen that a DELETE event does not clean up all resources, see #498 <#498>. I don't know if this happens in your case. To properly remove a Postgres cluster run kubectl delete postgresql postgres-postgres-operator-cluster. Then all elements incl. PDB, endpoints etc. should be removed. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#1057 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AP5SFQSKU6SROHE26RA7BDTR3RUMBANCNFSM4OYNGDPQ> .

[kahidna]

hi @FxKu I also tried do same (meaning reinit cluster with same name) but the error message still say
reason: 'Create' could not create cluster: could not create master endpoint: could not create master endpoint: endpoints \"acid-cluster-name" already exists.

even I reset the kubernetes cluster using kubeadm reset, delete the pv directory path, delete the docker images. the operator logs is same. Am I miss a step?

ps: my env :

• kubernetes 1.18.6
• postgres operator v1.5.0
• spilo 12 1.6-p3

[holgerson97]

I could fix this by delete both namespaces for operator and the desired cluster. When reapplying it still through that error. After 10min everything went to running. I then didn't ask questions anymore. I also had to go through that entire process a few times.