Upgrade of big instance takes time and consumes a lot of CPU for versions 6.4+

XMLWordPrintable

    • Severity 3 - Minor

      Problem:

      While upgrading big instance (~1500 plans, ~1500 deployments, ~7500 jobs) from 6.3 to 6.4/6.5 observing slowness and high CPU usage and constant records in logs like: 

      2018-05-10 11:47:06,927 DEBUG [AtlassianEvent::0-BAM::EVENTS:pool-1-thread-10] [DirtiedHibernateSessionObjectRecorder] flush dirty, scheduling cascaded invalidation in cache: com.atlassian.bamboo.hibernate.invalidators.PlanInvalidator@22e8ce29
2018-05-10 11:47:06,927 DEBUG [AtlassianEvent::0-BAM::EVENTS:pool-1-thread-10] [DirtiedHibernateSessionObjectRecorder] flush dirty, scheduling cascaded invalidation in cache: com.atlassian.bamboo.hibernate.invalidators.BuildDefinitionInvalidator@56a2db73

      After investigation was confirmed that ReencryptSecureFieldsUpgradeTask need to be fixed, as it processing all available tasks in all jobs.

      After some time (depends on instance size) CPU usage goes down, but issue is present again after the restart of Bamboo.

        1. bamboo-6.4-thread-dump.txt
          142 kB

            Assignee:
            Alexey Chystoprudov
            Reporter:
            Oleksandr Tkachenko
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated:
              Resolved: