Trying to add backups to existing cluster, `/run/etc/wal-e.d/env` is missing

[verdverm]

Please, answer some short questions which should help us to understand your problem / question better?

• Which image of the operator are you using? registry.opensource.zalan.do/acid/postgres-operator:v1.11.0
• Where do you run it? GKE
• Are you running Postgres Operator in production? yes
• Type of issue? question

I'm trying to add backups to an existing cluster, to push WAL-E to GCS.

I first updated the Helm Chart as follows (after doing the GCP IAM stuff)

helm upgrade --install \
  postgres \
  postgres-operator-charts/postgres-operator \
  --reuse-values \
  --set aws_or_gcp.wal_gs_bucket=<bucket-name> \
  --set podServiceAccount.name=postgres-pod-custom \
  --namespace=operators \
  --wait

Was hoping that setting aws_or_gcp.wal_gs_bucket would be sufficient

I've also tried adding a ConfigMap for the operator

Still, the /run/etc/... directory that should house the necessary wal-e.d stuff does not exist

root@psql-0:/home/postgres# ls -lh
total 8.0K
lrwxrwxrwx 1 root     root    8 Mar  6 13:19 etc -> /run/etc    <- DOES NOT EXIST
drwxr-xr-x 4 root     root 4.0K Mar 22 23:55 pgdata
-rw-rw-r-- 1 postgres root  156 Mar  6 13:08 pgq_ticker.ini
lrwxrwxrwx 1 root     root   17 Mar  6 13:19 postgres.yml -> /run/postgres.yml

Any help & guidance is much appreciated, thanks!

[verdverm]

Update, it seems this is the proper helm command

helm upgrade --install \          
  postgres \
  postgres-operator-charts/postgres-operator \
  --set configAwsOrGcp.wal_gs_bucket=<bucket name> \
  --set podServiceAccount.name=postgres-pod-custom \
  --namespace=operators \
  --wait

Note the format of --set configAwsOrGcp.wal_gs_bucket=<bucket name>

It seems this fills in the missing part, but now I am getting leader election issues. Does the instance fail to become healthy if it cannot connect to GCS via Workload Identity during startup?

[verdverm]

Ok, so it seems Workload Identity was the issue because the command changed

• OLD: https://github.com/zalando/postgres-operator/blob/master/docs/administrator.md#workload-identity-setup
• NEW: https://cloud.google.com/kubernetes-engine/docs/how-to/workload-identity#verify

[verdverm]

Looks to be working for me now.

I would say the only change needed would be to update the docs to reflect the new command format