[MAILBOX-405] Strengthen RENAME resiliency with retries

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 3.6.0
• Component/s: cassandra
• Assignee: Unassigned
• Reporter: Benoit Tellier
• Created: 03/Mar/21 04:01
• Updated: 11/Mar/21 05:06
• Resolved: 11/Mar/21 05:06

Description

Today one of my users did rename a folder containing ~25 folders. The RENAME did fail with a Cassandra level "Write timeout".

This resulted in an half renamed folder, harming Thunderbird experience (subscriptions were not updated)

We are investigating the cause of the write timeout. So far "reactor concurrency" do not loo like being to blame - but 16 concurrent renames is likely too much. I propose to decrease this parameter.

Also, positioning a retry would help...

Issue Links

• links to: GitHub Pull Request #319
• links to: GitHub Pull Request #324
• links to: GitHub Pull Request #326