Skip to content

Move meta resync into CLI command #3823

@roman-khimov

Description

@roman-khimov

Is your feature request related to a problem? Please describe.

I'm always frustrated when I think of metabase resynchronization. Currently it's done via resync_metabase configuration flag, but:

  • it requires configuration modifications to use
  • it's performed sequentially for shards which is slow (Parallel shard init #3416)
  • it's very easy to forget about it once resync is done since it can take quite a lot of time leading to unplanned resync on the next start
  • it's harder to track its state if something goes wrong since the node would just start normally

Describe the solution you'd like

  1. Drop configuration (it's not used normally, so not a problem to do this immediately).
  2. Move resync into lens meta resync which can take a blobstor and metabase file as parameters (don't forget about shard IDs).

This allows to run as many resynchronizations as needed in separate processes with clear result for each of them.

Metadata

Metadata

Assignees

No one assigned

    Labels

    I3Minimal impactS3Minimally significantU3RegularenhancementImproving existing functionalityneofs-lensNeoFS Lens app issuesneofs-storageStorage node application issues

    Type

    No type

    Projects

    No projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions