Hello, if I understand what you are after, I also had the need to include the cluster id in our backup paths (under azure);
( I wanted to make sure that if I needed to restore it the backups would go to a new location and have the backups "pristine" so I could attempt multiple restorations if need be).

I was able to accomplish this by including an initContainers section on the cluster that would query k8s for the postgresql cluster custom resource and store the uid in a configmap; then in the env section of the postgresql cluster I added a variable referencing the configmap / cluster id, finally, I overrode the WALG_AZ_PREFIX and referenced that newly added env var to get it into the path. I hope all this makes sense. (see below for more detail)

I was able to get this working by:

• Adding some additional resource/verbs to the "postgres-pod" cluster role:

            - apiGroups:
             - acid.zalan.do
             resources:
             - postgresqls
             verbs:
             - get
           - apiGroups: [""]
             resources: ["configmaps"]
             verbs:
             - get
             - create
             - update
             - patch

• Add an initContainers section to my postgresql cluster ( this initcontainer queries k8s to get your postgreql cr and retrieve / store the cluster's uid in a config map)

    initContainers:
    - args:
      - |
        pgc_id="$(kubectl get postgresql your-cluster-name -o jsonpath='{.metadata.uid}')";
        echo "{\"apiVersion\": \"v1\",\"kind\": \"ConfigMap\",\"metadata\": {\"name\": \"your-config-map-name\"}, \"data\": {\"pgc-id\": \"$pgc_id\"}}" > configmap.json;
        kubectl apply -f configmap.json;
      command:
      - /bin/sh
      - -c
      image: lachlanevenson/k8s-kubectl
      name: set-cluster-id-env

-Finally, I added an environment variable that is referenced as a dependent env var in the WALG_AZ_PREFIX (which I also include in the postgresql cluster to override it from the operator config):

    env:
      - name: PGC_ID
        valueFrom:
          configMapKeyRef:
            name: your-config-map-name
            key: pgc-id
      - name: WALG_AZ_PREFIX
        value: azure://your-blob-name/$(PGC_ID)/wal/$(PGVERSION)

I was able to overcome this using a solution similar to @mightymiracleman

In the Postgresql manifest

  env:
  - name: PG_UID
    valueFrom:
      fieldRef:
        fieldPath: metadata.uid

In the operator Pod config

WALG_AZ_PREFIX: azure://backups/spilo/$(SCOPE)/$(PG_UID)/wal

I was able to overcome this using a solution similar to @mightymiracleman

In the Postgresql manifest

  env:
  - name: PG_UID
    valueFrom:
      fieldRef:
        fieldPath: metadata.uid

In the operator Pod config

WALG_AZ_PREFIX: azure://backups/spilo/$(SCOPE)/$(PG_UID)/wal

I did give this a try as it is a lot cleaner than the way I was doing it; but I think this is pulling the uid from the POD and not the postgresql cluster. I added it to one of my clusters and was surprised to see different values for the 2 env vars I had; the first one was my postgresql cluster uid, the second one I tracked down to the pod that the spilo container was running in; I also bounced the pod and noticed the value changed (and matched up with the pod) did I miss something on that?

You are correct. I observed the same after a bit more testing. DownwardAPI doesn't support getting the UID from the StatefulSet or the Postgresql objects so I think your solution is the only way this will work.

No worries @youvegotmoxie ! I would have loved to do it the way you posted; it was a heck of a lot cleaner.

The intended capability of the operator (using the k8s dependent env-vars) doesn't work as we want to use it here, as the env list is ordered. The WAL_BUCKET_SCOPE_PREFIX and _SUFFIX envs are added as the last items and as such utilising them in the WALG_AZ_PREFIX doesn't work.

I'm trying to understand the logic in the code surrounding the ordering of the variables to see if I could create a PR for this.

Not sure what you are asking here (sorry); I added an init container to get the postgresql cluster id so I could set it in a config map (and reference it using the dependent env-var you mentioned).

I had to do this as we were using AKS / Azure blob storage; I don't think you have to do this with AWS. Apologies if this isn't what you were after.