Uploaded image for project: 'Jira Data Center'
  1. Jira Data Center
  2. JRASERVER-74911

User/group cache not fully replicated on node start.

XMLWordPrintable

      Issue Summary

      This is reproducible on Data Center: yes

      Steps to Reproduce

      1. Create a multinode (8) Jira DC environment
      2. Connect a remote directory with 100k users
      3. Start&stop node in cluster

      Expected Results

      All nodes have a consistent state of the user/group caches.

      Actual Results

      User may be missing in cache and Jira functionality based on this cache may not be able to find a user....

      Workaround

      1. Full cluster stop
      2. Start node and check if user/group caches state is as expected
      3. Only start new nodes if all nodes in current cluster has the correct state of the user/group caches

            [JRASERVER-74911] User/group cache not fully replicated on node start.

            Maciej Swinarski (Inactive) added a comment - - edited

            New logs:

            Example: Restoring 200k users in less than 4 sec:

            INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] Refreshing User cache from the database
INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] refreshing User cache from database, loaded 10000 Users
...
INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] refreshing User cache from database, loaded 80000 Users
...
INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] refreshing User cache from database, loaded 150000 Users
...
INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] refreshing User cache from database, loaded 200000 Users
INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] Completed refreshing User cache in PT3.942S

            Maciej Swinarski (Inactive) added a comment - - edited New logs: Example: Restoring 200k users in less than 4 sec: INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] Refreshing User cache from the database INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] refreshing User cache from database, loaded 10000 Users ... INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] refreshing User cache from database, loaded 80000 Users ... INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] refreshing User cache from database, loaded 150000 Users ... INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] refreshing User cache from database, loaded 200000 Users INFO [c.a.j.c.embedded.ofbiz.UserOrGroupCache] Completed refreshing User cache in PT3.942S

            Maciej Swinarski (Inactive) added a comment - - edited

            Monitoring cache replication related problems: 
            https://confluence.atlassian.com/enterprise/monitoring-the-cache-replication-954262834.html

            Maciej Swinarski (Inactive) added a comment - - edited Monitoring cache replication related problems:  https://confluence.atlassian.com/enterprise/monitoring-the-cache-replication-954262834.html

            Maciej Swinarski (Inactive) added a comment -

            From version 9.1 every node loads user/group caches (during node start) from DB and do not copy it from other nodes.
            Note that directory synchronisation is still propagated by sending cache replication messages with values. The change in behaviour is only applicable to how this cache is loaded on node start. 

            Maciej Swinarski (Inactive) added a comment - From version 9.1 every node loads user/group caches (during node start) from DB and do not copy it from other nodes. Note that directory synchronisation is still propagated by sending cache replication messages with values. The change in behaviour is only applicable to how this cache is loaded on node start. 

              mswinarski Maciej Swinarski (Inactive)
              mswinarski Maciej Swinarski (Inactive)
              Affected customers:
              0 This affects my team
              Watchers:
              3 Start watching this issue

                Created:
                Updated:
                Resolved: