Problem statement:

• Messages aren't in one place in Logstash, making feature-based dashboards harder to build.
• Shared dashboards used by deployers and SRE are harder to audit when the channel named after the component is missing messages.
• Quick summaries are more meaningful when messages aren't spread across multiple channels for the same component. E.g. a spike can more easily be identified this way, especially over long periods of time.
• Less mistakes or ambiguity in adding CA code, or in wmf-config.

Currently we have:

• CentralAuth - most messages
• CentralAuthVerbose - "on each page view" debug information containing personal names.
• CentralAuthRename - rename subcomponent diagnostics.
• suppressjob - Job queue job, CentralAuthSuppressUserJob.
• CentralAuthSULRename - CLI script, forceRenameUsers.php.

And configured as:

• CentralAuth - all severity levels enabled in prod, including "debug" and "info".
• CentralAuthVerbose - disabled in prod, enabled in beta cluster.
• CentralAuthRename - fully enabled in prod, including "debug" and "info".
• suppressjob - disabled.
• CentralAuthSULRename - disabled.

I propose the following:

• Rename CentralAuthRename to CentralAuth, any messages that should remain enabled by default in prod, elevate from to "info" level or above.
• Audit CentralAuth messages and elevate anything to "info" or above that we want enabled in prod.
• Rename CentralAuthVerbose to CentralAuth using debug level.
• Rename suppressjob and CentralAuthSULRename to CentralAuth using debug level.
• Simplify beta config as CentralAuth: debug and prod config as CentralAuth: info.