-
Bug
-
Resolution: Fixed
-
High
-
5.8.10, 5.10.4
-
18
-
Severity 3 - Minor
-
Summary
The internal administrator health check throws a false negative, most likely due to a high number of users in the internal directory.
Environment
An instance with a large number of users, e.g.: 17,000 or so.
Steps to Reproduce
- Startup Confluence with latest Confluence Instance Health.
- Ensure there are local admin users setup.
- Browse to the instance health page.
Expected Results
The health check passes, there are local admin users configured.
Actual Results
The health check fails, with a false positive.
Notes
I was able to reproduce this behavior by restoring a customer's backup. The internal directory, which is the only directory, contains 17,400+ users. The internal administrator health check fails, even though there are seven administrators set up, either through confluence-administrators membership or explicit global permissions. After removing all but the seven administrators, the health check shows a green result as expected.
Based on the code, the check works by loading up all of the internal directory users, and then iterating through the set to find administrators. There appears to be a maximum of 1000 set on the user entity query. Could this indeed be causing the false negative?
Temporary Workaround until bug is fixed
- Go to
> Add-ons - Click to expand the Confluence Healthcheck Plugin
- Click to expand the 29 of 29 modules on the right side
- Locate and disable the module named confluence.healthcheck.directory.internal.name (internalAdminCheck)
- relates to
-
-
- Closed
-
![[Extranet] Page](/images/icons/generic_link_16.png "[Extranet] Page"){kind=icon}
[CONFSERVER-45245] Internal administrator health check failing possibly due to high number of internal directory users
Hans-Peter Geier
added a comment - for customers on 5.10, what should they do? Ignore the error for now? Or are there manual patch instructions for older versions?
Hans-Peter Geier
added a comment - for customers on 5.10, what should they do? Ignore the error for now? Or are there manual patch instructions for older versions? A fix for this issue is now available for Confluence Server customers.
Upgrade now or check out the Release Notes to see what other issues are resolved.
This has been fixed in Confluence Instance Health 2.1.11, which can be updated from Manage add-ons or downloaded from the Atlassian Marketplace and uploaded to the Add-on Manager manually.
The fix was modifying one of the parameters in the hibernate query to be correctly passed as a String, rather than as an Enum. Additionally if the check fails to execute the hibernate query, we fail back to the previous method of executing the check.
The updated version is also planned to be bundled within Confluence 6.0.5.
hans-peter.geier For 5.10, they could update the plugin via marketplace. Please refer dcurrie@atlassian.com 's comment above https://jira.atlassian.com/browse/CONF-45245?focusedCommentId=1008268&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-1008268