Skip to content

Why is configuration split between global and type-specific stanzas? #3077

Open
@larsks

Description

@larsks

I wasn't sure if this counted as a bug or a feature request, so forgive the blank issue.

While configuring PGO for the first time (using version 5.0.4 of the operator in OpenShift), I'm confused about what options can be specified in the type-specific stanza under backups.pgbackrest.repos, and what options must go in backups.pgbackrest.global. For example, consider an S3 backup configuration. In backs.repos I have:

      - name: repo2
        schedules:
          full: ...
        s3:
          endpoint: "my-endpoint"
          bucket: netbox
          region: default

And I also have in backups.pgbackrest.global:

      global:
        repo2-s3-uri-style: path
        repo2-retention-full: "4"

Why can't these all put placed in the same section? E.g, all in backups.pgbackrest.repos:

      - name: repo2
        retention:
          full: "4"
        schedules:
          full: ...
        s3:
          endpoint: "my-endpoint"
          bucket: bucket
          region: default
          uri-style: path

Or alternately, all in backups.pgbackrest.global:

      global:
        repo2-retention-full: "4"
        repo2-s3-bucket: "bucket"
        repo2-s3-endpoint: "my-endpoint"
        repo2-s3-region: default
        repo2-s3-uri-style: path
        repo2-schedule-full: ...

In fact, this model would let one place the entire configuration into a ConfigMap and eschew both the pgbackrest.backups.global and the pgbackrest.backups.repos sections completely in favor of ConfigMap references in backups.pgbackrest.configuration.

Is that possible now? If not, what's the reason for splitting configuration options like this?

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions