@mattab opened this Issue on January 16th 2019 Member

In order to space disk size and costs, when a website is deleted, or when a segment is deleted, we need to make sure that archived reports stored in the archive_numeric* and archive_blob* tables are also deleted.

@tsteur commented on January 18th 2019 Member

refs https://github.com/matomo-org/matomo/issues/13976 when a user is deleted, we need to make sure to also delete all segments and their archives.

@mattab commented on February 20th 2019 Member

Notes from a call:

  • likely this could be implemented as a job that deletes all archives that are not part of the "Known segments"
  • for each archive table, get the list of unique done flags for each segment known in the system
    • use this list to delete all the other archives (whose done flags are not in this list)
  • making sure only archives older than 24 hours are deleted (as newer archives may be currently in use)
@tsteur commented on March 13th 2019 Member

@mattab how about segments that aren't automatically archived. I suppose we keep them in the archive tables? or delete them as well when they are older?

@tsteur commented on March 13th 2019 Member

@mattab are there maybe also range archives that we can delete?

@mattab commented on March 13th 2019 Member

Segments that aren't automatically archived, keep them forever.
Range archives might be already deleted automatically after some time, iirc. So there shouldn't need to do anything special with then.

Powered by GitHub Issue Mirror