Hello, we are running 14K flink applications. for some reason at a specific day we started to hit something that looks to be with the same exception stack of the issue described here:
https://issues.apache.org/jira/browse/FLINK-29427 affecting sporadically single digit % of the total deployments.
inside the link of the resolved bug report it seems this is fixed in flink 1.17. we are on 1.14.6 but this came out of nowhere for the first time after multiple years of operations. we are in an environment that infra changes and we are running on spot but none. the less nothing changed within the flink code we are using. is there any way of debugging such case? if i understand it some sort of instability?
---
update: after checking its not only janino related stack traces though this is the majority of them. but commong for everything is
trying to access closed classloader