Issue with User Synchronization Across Jira Data Center Nodes

[Becky]

Hi All,

We are running Jira Data Center and have encountered an issue we’ve never seen before.

One of our Jira admins created a user on a specific node. While the user information successfully propagated to two other nodes, it did not sync properly to the other two nodes.

On the nodes where the user was successfully created, all details—including full name, email, and directory—appear correctly, and the user is fully editable. However, on the node where the sync failed, the username is visible, but the full name, email, and directory information are missing, and it is not possible to edit the information for this user.

While we know that reindexing would resolve the issue, we’re trying to understand the root cause to prevent it from happening in the future.

Has anyone encountered this before, or does anyone have insights into what might have caused this and how to prevent it?

Any input would be greatly appreciated!

[Jim Knepley - ReleaseTEAM]

Hi @Becky 

You probably already suspect the same thing that I do, it sounds like it might be a cache replication problem.

It isn't easy to speculate without more details. If there were a network of system issue, I think the problems would be more widespread than just users. For example, you would have inconsistent results searching for issues based on which node is doing the search.

Cache replication logs are written to atlassian-jira.log; I would collect logs from each node to an aggregated view and watch for things like the droppedOnAddCounter and droppedOnSendCounter. The queueSize might also be interesting, but I would look there for slow performance, not failed replication.

(ref: https://confluence.atlassian.com/enterprise/monitoring-the-cache-replication-954262834.html)

I'm looking forward to Jira in the future, which will replace the current cache system with an optional OpenSearch server instance.