Problem Definition

Currently, when Jira Datacente node is shutdown while running FullReindex task (see related JRASERVER-47045, JRASERVER-66204), there is no way to remove this task from the cluster. All other nodes still think the task is running even the node is down, so this prevents corresponding operations to be done. 

Suggested Solution

Allow Jira system admin to delete stale FullReindex task.

Implemented Solution

• Before using this feature, please check if JRASERVER-68885 fits your needs as it's a much simpler and safer option.
• In order to remove stale re-index job from a cluster you can use REST API call (JIRA Sysadmin permission required):
  • Method: DELETE
  • URL: <JIRA>/rest/internal/2/reindex/unlock
• It will try to find active foreground re-index task and remove it if it belongs to OFFLINE node.
  • Since call removes only re-index task (which is assumed to be stuck), and doesn't kill the re-index thread itself, make sure that an actual node went down before you use this API (to ensure safety, there is also a check in the code).

Workaround

None