Uploaded image for project: 'Accumulo'
  1. Accumulo
  2. ACCUMULO-2441

Document internal state stored in RFile names


    • Type: Improvement
    • Status: Resolved
    • Priority: Major
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: 1.6.0
    • Component/s: tserver
    • Labels:


      A recent conversation on #Accumulo pointed out a need for better documentation on how we store state in RFile names

      15:18 < vines_> What are the prefix for files?
      15:18 < vines_> A is a full compaction, F is a partial major compaction, I is an import.. uhhh?
      15:20 < vines_> oops, F is minor, M is merging
      15:20 < elserj> C was manual compact i think?
      15:20 < vines_> I thought C was full
      15:21 < elserj> you just said A was full
      15:21 < vines_> I'm not sure
      15:21 < elserj> lol
      15:21 < vines_> considering I just contradicted myselkf on F
      15:21 < elserj> i thought A was automatic compact and C was user-issued compact
      15:21 < elserj> F was a flush, I thought
      15:22 < vines_> I don't think we differentiate user interfaced or not
      15:22 < vines_> C is a complete compaction - all files
      15:22 < vines_> A is an incomplete major compaction
      15:22 < vines_> F is a minor compaction
      15:22 < vines_> M is a merging minor
      15:22 < vines_> alright
      15:22 < elserj> ah ok. you'd typically see the C is a user did a compact in the shell
      15:22 < elserj> unless you set your ratio to something dumb?
      15:23 < elserj> that's probably what i was thinking of
      15:33 < vines_> perhaps
      15:43 < madrob> A is all.
      15:43 < madrob> C is partial
      15:43 < madrob> F is flush
      15:49 < busbey> could someone file a Jira to document this?
      15:49 < busbey> nevermind, I'll do it. ;)

      I'm not sure where the correct place for this is. I'm guessing somewhere in javadocs. Maybe in an advanced section of the administration manual?




            • Assignee:
              ecn Eric Newton
              busbey Sean Busbey
            • Votes:
              0 Vote for this issue
              2 Start watching this issue


              • Created: