Bug #115527 Contribution: Ensure FTS auxiliary tables are cleaned up properly
Submitted: 5 Jul 2024 14:35 Modified: 4 Dec 2024 9:52
Reporter: OCA Admin (OCA) Email Updates:
Status: Verified Impact on me:
None 
Category:MySQL Server: InnoDB storage engine Severity:S3 (Non-critical)
Version:8.0 OS:Any
Assigned to: CPU Architecture:Any
Tags: Contribution

[5 Jul 2024 14:35] OCA Admin
Description:
This bug tracks a contribution by Dirkjan Bussink (Github user: dbussink), as described in http://github.com/mysql/mysql-server/pull/554

How to repeat:
See description

Suggested fix:
See contribution code attached
[5 Jul 2024 14:35] OCA Admin
Contribution submitted via Github - Ensure FTS auxiliary tables are cleaned up properly 
(*) Contribution by Dirkjan Bussink (Github dbussink, mysql-server/pull/554#issuecomment-2209483395): I confirm the code being submitted is offered under the terms of the OCA, and that I am authorized to contribute it.

Contribution: git_patch_1953494582.txt (text/plain), 4.72 KiB.

[5 Jul 2024 16:05] Dirkjan Bussink
This affects at least MySQL 8.0 and all later versions.
[4 Dec 2024 9:52] MySQL Verification Team
Thank you for your contribution.
[26 Jul 11:21] Scorcia Luca Leonardo
Hi, I think I am affected by this bug. However this is having serious consequences on our platform which is a MySQL InnoDB cluster with Group Replication.

We have a long dropped database which had a FTS index that never got removed from the tablespace metadata. When we restart the first node, MySQL prints the MY-012351 warning but starts fine. However it will not rejoin the cluster automatically, and we have to bring it online manually using dba.rebootClusterFromCompleteOutage().

Worst, the second node of the cluster is simply unable to start anymore. We tried completely recreating it from scratch but after the cluster.addInstance command, using the Clone strategy, the node drops after data sync. MySQL startup on the second node fails with error "[MY-010020] [Server] Data Dictionary initialization failed.". We are unable to use the cluster anymore.

Is there a way to remove the orphaned FTS tablespaces from metadata?

Also, I noticed that the fix for this issue didn't get merged in trunk after a year. Can we get a status update?

Thanks in advance