Issue Summary

If Jira crashes during a rebalancing, the LOCK_HASH column keeps the lock, and it has to be manually cleaned, following https://confluence.atlassian.com/jirakb/ranking-failed-due-to-lock_hash-is-not-null-953658654.html.

Steps to Reproduce

1. Start Lexorank rebalancing
2. Kill Jira's process (simulating a crash)
3. Try to run a Lexorank balance

Expected Results

The balance works correctly.

Actual Results

We can observe the following error in the atlassian-greenhopper.log:

2020-12-21 15:11:31,668 lexorank-executor-thread-0 WARN      [service.lexorank.balance.LexoRankScheduledBalanceHandler] Balance for fieldId=10105 returned errors: Errors: 10105
Error Messages: [Jira Software cannot execute the rank operation at this time. Other users may be ranking the issues that you are trying to rank. Please try again later.]

Workaround

Follow the procedure described in https://confluence.atlassian.com/jirakb/ranking-failed-due-to-lock_hash-is-not-null-953658654.html.