Uploaded image for project: 'Ignite'
  1. Ignite
  2. IGNITE-12128

Potentially pds corruption on a failed node during checkpoint

    XMLWordPrintableJSON

    Details

    • Type: Bug
    • Status: Resolved
    • Priority: Critical
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: 2.7.6
    • Component/s: None
    • Labels:
      None
    • Release Note:
      Fixed an issue causing potential PDS corruption when a node is killed during checkpoint mark phase

      Description

      There are the case when we start a checkpoint but not create CP file marker, but PageMemory may start to flush dirty pages from checkpoint pages to page store. If node crashed at this moment, we can get inconsistency state, because we still not write checkpoint marker to disk but already write some pages for this checkpoint. If we try to recover from this state we cat get any sort of corruption problem. Recovery logic may not recognize that crash was during checkpoint because we did not write file marker when we start checkpoint but write some pages for this checkpoint.

        Attachments

          Issue Links

            Activity

              People

              • Assignee:
                akalashnikov Anton Kalashnikov
                Reporter:
                DmitriyGovorukhin Dmitriy Govorukhin
              • Votes:
                0 Vote for this issue
                Watchers:
                4 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 - 20m
                  20m