Description of problem:
When running the UI tests, we see 'Internal Server Error'.
ipactl status indicated Dir srv is not running.
Put debug in place, and rich is investigating
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Cannot recreate always
Created attachment 586071 [details]
The problem appears to be related to entryusn/preventryusn. The usn bepreop modifies the entry in the cache directly. If another thread doing a search happens to read the attribute list after preventryusn has been added by usn_bepreop_delete/_usn_add_next_usn, it will crash. I think the solution to this crash will be to _not_ modify the entry in the cache.
Was able to reproduce - it is very similar to https://bugzilla.redhat.com/show_bug.cgi?id=813964 - used almost the same steps except for step 1 - instead of setting up mmr, set up a single server with entryusn, referint, and memberof enabled. The rest of the steps are the same.
Technical note added. If any revisions are required, please edit the "Technical Notes" field
accordingly. All revisions will be proofread by the Engineering Content Services team.
Cause: Performing delete and search operations against the directory server under a high load with entryusn, memberof, and referential integrity enabled.
Consequence: Directory server crashes.
Fix: The entryusn code was modifying the entry in the cache directly. If a search thread was access the entry at the same time, the server would crash. The fix is to never modify the entry directly in the cache.
Result: Server does not crash when performing searches and deletions while under a high load with entryusn, memberof, and referential integrity enabled.
Ran same UI tests as above with ipa-server-2.2.0-16.el6.x86_64 and 389-ds-base-18.104.22.168-15.el6.x86_64, and didn't see the issue.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.