• Type: Bug
• Resolution: Duplicate
• Priority: High (View bug fix roadmap)
• Fix Version/s: Release 4.0, QA / Bug fixing
• Affects Version/s: Release 3.8
• Component/s: None
• Labels:

  • affects-server
• Environment:

  Java 1.6.0 update 16

[JSWSERVER-1465] Greenhopper can crash 64bit JVMs under Linux

Collapse comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:18 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:18 AM

Duplicate of http://jira.atlassian.com/browse/GHS-1357

The patch provided there should fix the problem if you are affected by this issue.

Expand comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:18 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:18 AM Duplicate of http://jira.atlassian.com/browse/GHS-1357 The patch provided there should fix the problem if you are affected by this issue.

Collapse comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:13 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:13 AM

Patch on http://jira.atlassian.com/browse/GHS-1357 seems to fix this

Expand comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:13 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:13 AM Patch on http://jira.atlassian.com/browse/GHS-1357 seems to fix this

Collapse comment: JC added a comment - 09/Sep/2009 5:12 AM

JC added a comment - 09/Sep/2009 5:12 AM

Yeah ok this looks like the GHS-1357 issue.
GH tries to reassign a ranking to the sub but it ends into a death loop.

The patch (3.8.1) should be correcting this.

Let me know.

Cheers,

Expand comment: JC added a comment - 09/Sep/2009 5:12 AM

JC added a comment - 09/Sep/2009 5:12 AM Yeah ok this looks like the GHS-1357 issue. GH tries to reassign a ranking to the sub but it ends into a death loop. The patch (3.8.1) should be correcting this. Let me know. Cheers,

Collapse comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:09 AM, Edited by Andrew Myers [Atlassian] - 09/Sep/2009 5:12 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:09 AM - edited

Edit: I could actually reproduce it on Java 1.6.0 update 10 and i guess i should have refeshed the issue

Expand comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:09 AM, Edited by Andrew Myers [Atlassian] - 09/Sep/2009 5:12 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:09 AM - edited Edit: I could actually reproduce it on Java 1.6.0 update 10 and i guess i should have refeshed the issue

Collapse comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:07 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:07 AM

Managed to replicate this without customer data.

The problem is that there is an issue with multiple subtasks which are displaying "Issue is not rankable" in the ranking custom field. I did this by creating some subtasks for an issue, assign them a fix version, then release and archive that particular version. Then I added a ranking custom field and enabled it for the project.

What seems to be happening is that when say one of the subtasks is reopened the GreenHopperCTFIndexer is run. During that process it tries to reorder the subtasks ranking, but because there are multiple tasks with a null ranking field, it seems to bounce between the issues trying to reorder them. Eventually something happens the JVM just gives up the ghost and apparently segfaults.

Expand comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:07 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 5:07 AM Managed to replicate this without customer data. The problem is that there is an issue with multiple subtasks which are displaying "Issue is not rankable" in the ranking custom field. I did this by creating some subtasks for an issue, assign them a fix version, then release and archive that particular version. Then I added a ranking custom field and enabled it for the project. What seems to be happening is that when say one of the subtasks is reopened the GreenHopperCTFIndexer is run. During that process it tries to reorder the subtasks ranking, but because there are multiple tasks with a null ranking field, it seems to bounce between the issues trying to reorder them. Eventually something happens the JVM just gives up the ghost and apparently segfaults.

Collapse comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 4:51 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 4:51 AM

Incorrectly set the fix version instead of the affects version.

Expand comment: Andrew Myers [Atlassian] added a comment - 09/Sep/2009 4:51 AM

Andrew Myers [Atlassian] added a comment - 09/Sep/2009 4:51 AM Incorrectly set the fix version instead of the affects version.

Collapse comment: JC added a comment - 09/Sep/2009 3:41 AM, Edited by JC - 09/Sep/2009 5:04 AM

JC added a comment - 09/Sep/2009 3:41 AM - edited

Andrew can you see if GHS-1357 fixes the problem on the customer data?
The patch is in the comments.

It might be caused by the stackoverflow.

Cheers,
JC

Expand comment: JC added a comment - 09/Sep/2009 3:41 AM, Edited by JC - 09/Sep/2009 5:04 AM

JC added a comment - 09/Sep/2009 3:41 AM - edited Andrew can you see if GHS-1357 fixes the problem on the customer data? The patch is in the comments. It might be caused by the stackoverflow. Cheers, JC