Clusters aren't working properly when created with 1 replica

[eliranw]

Overview

I create 2 clusters with a helm chart, using a yaml template. I create users using the users: in the yaml in both times. I found out that the cluster with 1 replica doesn't create the users after the creation. I checked this by connecting to the postgres pod and running a \du. So I tried changing variables around and deleting and recreating the postgrecluster resource, and only when creating the cluster with 2 replicas I found out that the users are being created. I also found out that these clusters with the single replica aren't being scaled up when I change the yaml to be 2 replicas after I initially created the cluster with 1. I checked in the postgresclusters resource and the changes are happening there, but I guess the operator isn't triggering any changes to occur to the actual postgres/statefulset.

I looked in the pgdata/<Cluster Name>/logs directory but didn't find any logs referencing the users that aren't being created. I also looked at the operator logs and didn't find anything.

I am currently running on k3s version v1.22.6+k3s1, but this also happened to me on an AKS cluster.

I also think this might be something with synchronous replication configured, as I see that also when creating a user with psql commands, they hang.

apiVersion: postgres-operator.crunchydata.com/v1beta1
kind: PostgresCluster
metadata:
  name: REDACTED
spec:
  patroni:
    dynamicConfiguration:
      synchronous_mode: true
      synchronous_mode_strict: true
  image: registry.developers.crunchydata.com/crunchydata/crunchy-postgres:centos8-13.5-0
  postgresVersion: 13
  users:
  - databases:
    - REDACTED
    - postgres
    name: REDACTED
    options: SUPERUSER
  - databases:
    - REDACTED
    - REDACTED
    name: REDACTED
    options: SUPERUSER
  databaseInitSQL:
    name: REDACTED-db-init
    key: REDACTED_init.sql
  instances:
    - name: instance1
      replicas: 2
      dataVolumeClaimSpec:
        accessModes:
        - "ReadWriteOnce"
        resources:
          requests:
            storage: 500Gi`

Thanks for reporting the issue, we're looking forward to helping you!

Add a concise description of what the bug is.

Environment

Please provide the following details:

• Platform: Rancher K3S/AKS
• Platform Version: Rancher- v1.22.6+k3s1,
• PGO Image Tag: (e.g. ubi8-5.1.0-0) centos8-13.5-0
• Postgres Version (e.g. 14) 13
• Storage: hostpath

Steps to Reproduce

1. Create cluster with a single replica and a users: configuration as shown in the yaml I sent
2. Connect to postgres cluster and run \du
3. See that user isn't being created.

EXPECTED

1. Users are being created normally.

ACTUAL

1. Users arent created

[jmckulk]

hey @eliranw, what version of the operator are you using? Can you share the image tag from your operator deployment?

[tjmoore4]

Closing this issue. Please reopen if you are still encountering this problem and would like further assistance.