High CPU load at specific time for few hours
Problem
Daily, at a specific time, for example, 1 am, the CPU load is higher than 50 % most of the time.
The following appears in the atlassian-confluence.log
2015-08-07 01:11:00,012 INFO [scheduler_Worker-1] [userprofile.synchronization.modules.LDAPUserDataSyncJob] doExecute Starting job based synchronization for all LDAP users.
...
2015-08-07 03:03:43,366 WARN [scheduler_Worker-1] [confluence.util.profiling.DurationThresholdWarningTimingHelperFactory] logMessage Execution time for publishing event de.communardo.confluence.plugins.userprofile.synchronization.modules.LDAPUserDataSyncFinishedEvent[source=de.communardo.confluence.plugins.userprofile.synchronization.modules.LDAPUserDataSyncJob@3414c2d5] took 2735638 ms (warning threshold is 5000 ms)
2015-08-07 03:03:43,367 INFO [scheduler_Worker-1] [userprofile.synchronization.modules.LDAPUserDataSyncJob] doExecute Finished job based synchronization for all LDAP users.Additionally, a separate problem is if the synchronization fails with a log message similar to this:
... [plugins.userprofile.export.UserProfileExporterService] doInternalExport Error during user profile export
java.io.IOException: Unable to create the final export file
at de.communardo.confluence.plugins.userprofile.export.UserProfileExporterService.doInternalExport(UserProfileExporterService.java:226)Cause
This behavior is caused by a CSV export that is automatically generated after the nightly synchronization of the User Profiles for the Confluence app.
Solution/Workaround
If you do not need the automatic nightly export, there is a workaround to mitigate that problem:
Go to the user data export configuration via Administration > Users & Security - User Profile Configuration > Export
At "Filter", enter a phrase that will filter out every user for the export
example phrase "filterWhichCausesExportToBeEmpty"
Click "Save"
That will cause the nightly export to finish instantly and prevent a high load over a longer period.
If you need the export or assistance with the workaround, please contact our support team.