Issue Summary

When we start the Reindex process using the REST API /rest/api/2/reindex?type=FOREGROUND, the index snapshot is still able to run on other nodes, which results in corrupt snapshots.

Steps to Reproduce

1. Start both JIRA nodes
2. Schedule the index snapshot to run
3. Initiate full locked reindex at Node1 a few minutes before the snapshot:

   curl  -X POST http://127.0.0.1:8091/rest/api/2/reindex?type=FOREGROUND
   

4. Notice that the snapshot still happens on Node2

Expected Results

The index snapshot should not run while the reindex is running.
It should either queue it up and wait for the reindex to finish, or simply handle as a conflict and not run the snapshot.

Actual Results

The index snapshot runs while the reindex is running.

Workaround

• When calling the reindex, make sure that the snapshot is not scheduled to run in the next minutes/hours
• Or make sure to start it from the UI, instead of the API.