Uploaded image for project: 'Accumulo'
  1. Accumulo
  2. ACCUMULO-4575

Concurrent table delete operations leave orphan fate transaction locks

    Details

    • Type: Bug
    • Status: Resolved
    • Priority: Minor
    • Resolution: Fixed
    • Affects Version/s: 1.7.2
    • Fix Version/s: 1.7.3, 1.8.1, 2.0.0
    • Component/s: fate, master
    • Labels:
      None

      Description

      There is an issue where orphan zookeeper transaction locks (as reported by the FATE print command) are being created if delete commands are issued for the same table concurrently.

      The DELETE fate transaction does not keep track of the table id and when one of the delete operations succeeds, any others queued throw an exception because the table does not exist any more - this causes the existing zookeeper transaction locks to become orphaned and they required manual clean-up.

      This may be related to ACCUMULO-4553.

        Issue Links

          Activity

          Hide
          kturner Keith Turner added a comment -

          While working on this issue, I have come the conclusion that it is a broader problems. Its not just concurrent delete table operations, its fate operations in general running concurrently with a delete table op. I think a compact operation started while a delete table op is running could also have problems getting the namespace id. I think all FATE operations should resolve the namespace id once before starting the fate op.

          Show
          kturner Keith Turner added a comment - While working on this issue, I have come the conclusion that it is a broader problems. Its not just concurrent delete table operations, its fate operations in general running concurrently with a delete table op. I think a compact operation started while a delete table op is running could also have problems getting the namespace id. I think all FATE operations should resolve the namespace id once before starting the fate op.
          Hide
          mjwall Michael Wall added a comment -

          Moved back to 1.8.1 as Keith Turner found other stuff to do. We should either close this and create a new ticket or finish that work before the release.

          Show
          mjwall Michael Wall added a comment - Moved back to 1.8.1 as Keith Turner found other stuff to do. We should either close this and create a new ticket or finish that work before the release.

            People

            • Assignee:
              kturner Keith Turner
              Reporter:
              etcoleman Ed Coleman
            • Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:

                Time Tracking

                Estimated:
                Original Estimate - Not Specified
                Not Specified
                Remaining:
                Remaining Estimate - 0h
                0h
                Logged:
                Time Spent - 4h 10m
                4h 10m

                  Development