No, I didn’t. But I tried to rise the lock timeout to 300 seconds and the problem persisted
The point is that the same code runs without any problem on 7.3 and 7.2.
My intuition is that there is something to fine tune on the 7.4 images, so I would like to know if there is somebody else with this issue, and then try to escalate it.
I’ve never run over such an issue. Checked: The Mysql Server is the same (compared with the 7.3 build-> 5.7.29-0ubuntu0.18.04.1) maybe diffing the PHP mysqli configuration between 7.3 and 7.4?
My suggestion with restarting the transaction was not that resourceful as I’ve now seen it is within a migration script. If for some reason the query is different though, it might be worth to log it in full.
Have you tried running it on a different system than Travis with that Mysql and PHP version?