Closed
Description
Discussed in #3007
Originally posted by raul-avila-ph March 10, 2025
As we explained in #3006, we had some issues trying to read the count
field in x-death
header. We have proposed a safer way to parse it.
When we had the problem in production, it took some time to be detected, as the ClassCastException was logged at DEBUG level:
In this discussion we'd like to ask you if it'd be possible for the framework to log this kind of errors at the WARN or ERROR level, maybe filtering out exceptions that shouldn't create unnecessary noise. The fact that a ClassCastException was only logged at DEBUG level made it invisible for our monitoring systems and it delayed the process to diagnose the problem.