Following an upgrade of our AWS RDS instance from MySQL version 5.7.42 to version 8.0.36, we are encountering persistent deadlock issues across concurrent transactions, affecting even basic read operations within our Django 4.1 application. Notably, these deadlocks were not present prior to the upgrade in version 5.7. It's important to highlight that the error message received is: Deadlock found when trying to get lock; try restarting transaction.
Interestingly, attempts to diagnose the issue through the InnoDB monitor output have been unsuccessful, as these specific deadlock incidents do not appear in the output. However, when inducing a conventional deadlock scenario intentionally, the expected monitor output is observed, suggesting that the deadlocks we are experiencing with the application are atypical.
Efforts to resolve the issue through various configurations adjustments, including changing isolation levels and modifying parameter group parameters related to deadlock detection and logging, have not yielded success.
We are in the process of implementing Percona monitoring tools with the aim of capturing extensive logs and information that might facilitate the diagnosis of these deadlock issues.
Any insights or suggestions on how to address and resolve these deadlock issues would be greatly appreciated.
