pt-table-checksum - rollback() due to DESTROY without explicit disconnect
General
Escalation
General
Escalation
Description
Since version 3.5.1 we're getting multiple errors when running pt-table-checksum like:
This seems to be caused by PT-2138 wait_for_slaves change:
We've reverted that and got rid of the error:
Any idea why we're getting this?
Environment
CentOS 7, MariaDB 10.4.28
Activity
Show:
Simon Gauthier May 6, 2024 at 12:28 PM
Hi,
The bug is still present in percona-toolkit 3.5.7, on a primary MariaDB 10.5.22 and a replica MariaDB 10.5.23.
matthew December 28, 2023 at 3:33 PM
I should also note that I’m running the newest percona mysql server 8.0.x version (8.0.34-26) and toolkit versions (3.5.7)
matthew December 28, 2023 at 3:32 PM
I’m experiencing this issue and had reported a new bug which was marked as a duplicate. I’m running the above statement on the primary (read/write, myprimarydbhost). It runs as the root user and my .my.cnf is configured with the root user/pass. It automatically discovers the single secondary (read only) server and connects using the same root user/pass defined in the .my.cnf.
Aaditya Dubey November 27, 2023 at 10:35 AM
Hi ,
Sure, Thank You!
Sarvesh Goburdhun November 27, 2023 at 9:33 AM
Hello all,
I did notice there is a new version of percona-toolkit out.
l'll update it on our servers while waiting for your update on the issue.
On my side, I'll inform if the issue continues to occur.
Since version 3.5.1 we're getting multiple errors when running pt-table-checksum like:
This seems to be caused by PT-2138 wait_for_slaves change:
We've reverted that and got rid of the error:
Any idea why we're getting this?