Uploaded image for project: 'HPCC'
  1. HPCC
  2. HPCC-16839

Dali transaction log could go out of sync, if backup node down and Dali shutdown abruptly.


    • Type: Bug
    • Status: Resolved
    • Priority: Major
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: 6.2.6
    • Component/s: Dali
    • Labels:


      Description of problem from James Wiltshire :

      A new theory...

      I mentioned that after the "event", Dali ends up with massive extra data in the delta/inc/det file.

      Turns out, in this last instance - there were phases to the event -

      1. Dali lost connectivity to the backup server

      2. Dali server went down

      I mentioned there over 8000+ apparently-redundant chunks in the delta file.

      And today I noticed there many (8000+) "saveDelta" messages in the Dali log file.

      What I believe may be happening:

      • Dali attempts to write the delta file
      • It fails writing to backup (in this case, because network connectivity dropped)
      • Accordingly, it does not mark the delta info as "saved"
      • But it still writes the entire delta to the local delta file

      The effect of this:

      The local delta file has updates that it has already received and doesn't "need".

      The backup delta file "needs" them, but not the local delta file.

      So - if this is the case...

      Then simply losing network connectivity to the backup server could "corrupt" the local delta file.




            • Assignee:
              jakesmith Jake Smith
              jakesmith Jake Smith
            • Votes:
              0 Vote for this issue
              1 Start watching this issue


              • Created: