initContainer defaults preventing existing YAML from working

[aquarapid]

I have a (test) operator cluster YAML that has a keyspace section looking like this:

  keyspaces:
  - name: main
    partitionings:
    - equal:
        parts: 2
        shardTemplate:
          databaseInitScriptSecret:
            name: example-cluster-config
            key: init_db.sql
          replication:
            enforceSemiSync: true
          tabletPools:
          - cell: uscentral1a
            type: replica
            replicas: 1
            vttablet:
              extraFlags:
                db_charset: utf8mb4
              resources:
                requests:
                  cpu: 10m
                  memory: 256Mi
            mysqld:
              resources:
                requests:
                  cpu: 10m
                  memory: 256Mi
            dataVolumeClaimTemplate:
              accessModes: ["ReadWriteOnce"]
              resources:
                requests:
                  storage: 5Gi
              storageClassName: uscentral1a-ssd
          - cell: uscentral1c
            type: replica
            replicas: 1
            vttablet:
              extraFlags:
                db_charset: utf8mb4
              resources:
                requests:
                  cpu: 10m
                  memory: 256Mi
            mysqld:
              resources:
                requests:
                  cpu: 10m
                  memory: 256Mi
            dataVolumeClaimTemplate:
              accessModes: ["ReadWriteOnce"]
              resources:
                requests:
                  storage: 5Gi
              storageClassName: uscentral1c-ssd
          - cell: uscentral1f
            type: replica
            replicas: 1
            vttablet:
              extraFlags:
                db_charset: utf8mb4
              resources:
                requests:
                  cpu: 10m
                  memory: 256Mi
            mysqld:
              resources:
                requests:
                  cpu: 10m
                  memory: 256Mi
            dataVolumeClaimTemplate:
              accessModes: ["ReadWriteOnce"]
              resources:
                requests:
                  storage: 5Gi
              storageClassName: uscentral1f-ssd

This used to work just fine. But since PR #216 I get the multiple errors (one for each vttablet) when applying this (can be seen with kubectl get events or in the operator logs):

2m38s       Warning   CreateFailed             vitessshard/example-main-x-80-84bc4397                                   failed to create Pod example-vttablet-uscentral1f-2191363457-7dead7ff: Pod "example-vttablet-uscentral1f-2191363457-7dead7ff" is invalid: [spec.initContainers[0].resources.requests: Invalid value: "256Mi": must be less than or equal to memory limit, spec.initContainers[1].resources.requests: Invalid value: "256Mi": must be less than or equal to memory limit]

I can resolve this in one of two ways:

• Reverting Set vttablet init container resource limits #216
• Reducing the memory requests for the vttablets to 128Mi. However, this isn't a real solution, since it would pack too many vttablet pods into a node where I want to run fewer/larger vttablets...

Environment is GKE (1.19.16-gke.1500).