Unreliable Issue Navigator

[Nejc Grenc]

Background

Recently we have made the following changes to our Jira server:

• upgraded Jira from version 7.8.1 to version 8.1.0
• upgraded our custom Jira plugin to version 8.1.0 respectivelly
• installed Portfolio for Jira

The changes made were successfully implemented. However, since then we have been experiencing weird behaviour with Issue Navigator.


Issue

The Issue Navigator is unreliably executing queries. It is returning random and inconsistent amount of results.
The following two images are 3 seconds apart and I can assure you that no changes have been to any of the issues involved in this query. This is only one example of many.

Suspicion

We are suspecting this is related to caching mechanism

com.atlassian.cache.*

as the issue starts occurring about 30 minutes after every server restart. And the only "timer" we have set up is the cache expiration period (30 minutes).

We are currently trying to remove

CacheManage

usage, but we would very much rather fix the caching mechanism instead of tossing it aside.

Is there a known bug that relates to our issue?
Any help and ideas from the community are very welcome.

[Earl McCutcheon]

Hi Nejc,

It looks like something is going on with the images you linked from the external site, and they are not showing up for us.  Can you try uploading them in your comment directly so we can take a look, this can be done via the camera icon in the comment box and when you drag'n'drop the image into the popup, use the Upload tab rather than the URL tab, and upload the images as a png or jpeg.

But from the behavior you are describing it does sound like a caching issues on the surface so I would say you are on the right track for troubleshooting

A method to verify if this is cache corruption is to purge the add-on and application caching from the server filesystem with a restart and lock and re-index after the restart.  This will give you a purge and full rebuild on any caching on the app's, the bundled plugins, and the application, Noting that if this is a large production site a full lock and re-index will lock out all users until the index rebuild is complete so if a manual re-index takes a while to complete make sure to schedule downtime accordingly, the method to accomplish this is:

1. Stop Jira
2. Remove the hidden cache files ".bundled-plugins" and ".osgi-plugins" manually in the $jira-home/plugins directory

   $jira-home/plugins/.bundled-plugins
   
   $jira-home/plugins/.osgi-plugins

3. Delete the contents of the work directory located in the $jira-install/ directory (!)NOTE: Only remove the contents of the work directory, do not remove the work directory path directly, or the path will not rebuild without manually creating the file path and setting permissions

   $jira-install/work/*

4. Delete the contents of the Cache directory, again contents only not the base directory:

   $jira-home/caches/*

5. Restart your JIRA instance and watch the atlassian-jira.log file for anything else that may be occuring on the backend during startup that could help narrow down any conflicts
6. Once the application is running trigger a full lock and re-index (The issue data will not be visible until this completes)

Regards,
Earl