AccumulatingResultBuilder needs to provide a way to get full stack traces during reindex

XMLWordPrintable

    • Type: Bug
    • Resolution: Fixed
    • Priority: Medium
    • 6.2-OD-5
    • Affects Version/s: 6.1.3
    • Component/s: None
    • 6.01

      As discovered while troubleshooting reindexing problems on an internal JIRA instance, exceptions during a reindex are caught and traced as one-liners:

                      try
                {
                    result.await();
                    successes++;
                }
                catch (RuntimeException e)
                {
                    failures++;
                    log.warn(e.getMessage());
                }

      When several issues failed to reindex due to a problem with a custom field, the result was a string of warnings like this (edited):

      2013-11-15T06:23:02.135046+00:00 hostname: 2013-11-15 00:23:02,134 JiraTaskExectionThread-1 WARN admin 22x79x1 14p2q1a 127.0.0.1 /secure/admin/IndexReIndex.jspa [atlassian.jira.index.AccumulatingResultBuilder] java.util.concurrent.ExecutionException: java.lang.NullPointerException
2013-11-15T06:23:02.137881+00:00 hostname: 2013-11-15 00:23:02,137 JiraTaskExectionThread-1 WARN admin 22x79x1 14p2q1a 127.0.0.1 /secure/admin/IndexReIndex.jspa [atlassian.jira.index.AccumulatingResultBuilder] java.util.concurrent.ExecutionException: java.lang.NullPointerException
2013-11-15T06:23:02.138059+00:00 hostname: 2013-11-15 00:23:02,138 JiraTaskExectionThread-1 WARN admin 22x79x1 14p2q1a 127.0.0.1 /secure/admin/IndexReIndex.jspa [atlassian.jira.index.AccumulatingResultBuilder] java.util.concurrent.ExecutionException: java.lang.NullPointerException

      While this is understandable, since the result is that the problem issue does not get indexed, there needs to be a way to view a) which issue had the problem and b) a full stack trace.

      In my opinion, it is ok to require DEBUG logging to be enabled before we dump the stack trace, but others may feel differently. I think the issue whose indexing failed needs to be reported either way.

            Assignee:
            Trevor Campbell (Inactive)
            Reporter:
            crf
            Votes:
            0 Vote for this issue
            Watchers:
            9 Start watching this issue

              Created:
              Updated:
              Resolved: