Uploaded image for project: 'Jira Software Data Center'
  1. Jira Software Data Center
  2. JSWSERVER-16473

Reindex All FAILED - Caused by null issuenum entry in jiraissue table

XMLWordPrintable

    • 7.02
    • 80
    • Severity 2 - Major
    • 229
    • Hide

      Hi all,

      I'm glad to announce that Jira versions 8.9.1, 8.10.0, and higher come with a sanity check that will prevent Jira from trying to index such corrupted issue entries.

      While this doesn't fix or prevent the corrupted data in the jiraissue table (see JRASERVER-69826 for that), it prevents the index corruption.

      The sanity check was only added to the latest Jira versions because Jira 8.9.0 comes with a huge rewrite of the affected classes. However, this fix alone is quite small and is a likely candidate for backporting to the 8.5 ER.

      Daniel Rauf

      Show
      Hi all, I'm glad to announce that Jira versions 8.9.1, 8.10.0, and higher come with a sanity check that will prevent Jira from trying to index such corrupted issue entries. While this doesn't fix or prevent the corrupted data in the jiraissue table (see JRASERVER-69826 for that), it prevents the index corruption. The sanity check was only added to the latest Jira versions because Jira 8.9.0 comes with a huge rewrite of the affected classes. However, this fix alone is quite small and is a likely candidate for backporting to the 8.5 ER. Daniel Rauf

      Summary

      Under a specific scenario, when doing a Bulk Change -> Edit Issues, Jira ends up with a NULL value under the issuenum column of jiraissue table, causing indexing failures with Reindex All FAILED error.

      Messages like the one below appear in the logs during reindex:

      2020-04-16 23:21:15,522+1000 JiraTaskExecutionThread-2 WARN admin 1401x13091x1 1jldanz 127.0.0.1 /secure/admin/IndexReIndex!reindex.jspa [c.a.jira.index.AccumulatingResultBuilder] com.atlassian.jira.util.dbc.Assertions$NullArgumentException: parameter should not be null!
java.lang.RuntimeException: com.atlassian.jira.util.dbc.Assertions$NullArgumentException: parameter should not be null!
	at com.atlassian.jira.index.DefaultIndex$Failure.<init>(DefaultIndex.java:100)
	at com.atlassian.jira.issue.index.DefaultIssueIndexer$IndexIssuesOperation.perform(DefaultIssueIndexer.java:469)
	at com.atlassian.jira.issue.index.DefaultIssueIndexer$IndexIssuesOperation.perform(DefaultIssueIndexer.java:438)
	at com.atlassian.jira.issue.index.DefaultIssueIndexer.lambda$null$2(DefaultIssueIndexer.java:365)
	at com.atlassian.jira.index.SimpleIndexingStrategy.apply(SimpleIndexingStrategy.java:7)
	at com.atlassian.jira.index.SimpleIndexingStrategy.apply(SimpleIndexingStrategy.java:5)
	at com.atlassian.jira.index.MultiThreadedIndexingStrategy$1.call(MultiThreadedIndexingStrategy.java:33)
	at com.atlassian.jira.index.MultiThreadedIndexingStrategy$1.call(MultiThreadedIndexingStrategy.java:31)
	at com.atlassian.jira.util.concurrent.BoundedExecutor$2.call(BoundedExecutor.java:68)
	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
	at java.lang.Thread.run(Thread.java:748)
Caused by: com.atlassian.jira.util.dbc.Assertions$NullArgumentException: parameter should not be null!
	at com.atlassian.jira.util.dbc.Assertions.notNull(Assertions.java:32)
	at com.atlassian.jira.issue.changehistory.ChangeHistoryItem$Builder.inProject(ChangeHistoryItem.java:326)
	at com.atlassian.jira.issue.history.AbstractDateRangeBuilder.changeItemBuilder(AbstractDateRangeBuilder.java:95)
	at com.atlassian.jira.issue.history.VersionDateRangeBuilder.createInitialChangeItem(VersionDateRangeBuilder.java:69)
	at com.atlassian.jira.issue.history.VersionDateRangeBuilder.buildDateRanges(VersionDateRangeBuilder.java:110)
	at com.atlassian.jira.issue.index.DefaultChangeHistoryRetriever.filterChangeItems(DefaultChangeHistoryRetriever.java:74)
	at com.atlassian.jira.issue.index.DefaultChangeHistoryRetriever.createChangeGroupsFromChangeItems(DefaultChangeHistoryRetriever.java:47)
	at com.atlassian.jira.issue.index.DefaultChangeHistoryRetriever.apply(DefaultChangeHistoryRetriever.java:41)
	at com.atlassian.jira.issue.index.DefaultChangeHistoryRetriever.apply(DefaultChangeHistoryRetriever.java:25)
	at com.atlassian.jira.issue.index.DefaultIssueIndexer$DefaultDocumentCreationStrategy$1.apply(DefaultIssueIndexer.java:626)
	at com.atlassian.jira.issue.index.DefaultIssueIndexer$DefaultDocumentCreationStrategy$1.apply(DefaultIssueIndexer.java:623)
	at com.atlassian.jira.issue.index.DefaultIssueIndexer$DefaultDocumentCreationStrategy.get(DefaultIssueIndexer.java:616)
	at com.atlassian.jira.issue.index.DefaultIssueIndexer$IndexIssuesOperation.perform(DefaultIssueIndexer.java:450)
	... 11 more

      Steps to Reproduce

      1. Create a clean Jira installation;
      2. Create a Jira project;
      3. Create 2+ issues;
      4. Goto Issues -> Search for issues;
      5. On a new browser window or tab, open one of the recently created issues;
      6. Back to the original window or tab (search for issues page), click the Tools button and choose Bulk Change: all 2+ issue(s);
      7. Select all 2+ and click next;
      8. Select Edit Issues and click next;
      9. At this point, delete the issue you opened in a different window or tab;
      10. Go back to the original window or tab, edit something, click next and confirm the change;
      11. Wait for the operation to complete;
      12. Run the following query, a record will be returned with almost all field set as null. 
        SELECT * FROM jiraissue WHERE issuenum IS NULL;

      Expected Results

      It was expected to have the deleted record/row completely gone from jiraissue table, and only the remaining records updated, or maybe have Jira showing a friendly error message to the user.

      Actual Results

      After the bulk operation is 100% complete, the deleted record remains in the table jiraissue, but with null values for all fields, with exception of the id and updated fields. Fields updated in the bulk process may also have a value different from NULL 

      Workaround

      1. Remove the NULL entries
        1. Stop Jira;
        2. Perform a complete database backup in case you need to rollback;
        3. Run the following query: 
          DELETE FROM jiraissue WHERE issuenum IS NULL;
        4. Start Jira;
      2. Do a full Re-index, by choosing the option to 'Lock and Re-Index'.

              drauf Daniel Rauf
              amacedo@atlassian.com Alessandro Macedo
              Votes:
              59 Vote for this issue
              Watchers:
              74 Start watching this issue

                Created:
                Updated:
                Resolved: