Pod "CrashLoopBackOff" after upgrade to 1.66.2

[pollynesterovich]

After trying to upgrade vmagent from version 1.63.0 to 1.66.2, the pod crashes constantly "CrashLoopBackOff". There are no errors in the logs, the pod dies due to oom. Optimal memory and processor limits set. There were no such problems before the upgrade. Please tell me what is the reason for the increase in resources for consumption and how to fix it in order to switch to the new version of vmagent?

[f41gh7]

Can you temporary increase memory limit until it doesn't crash. Then collect memory profile according to doc https://docs.victoriametrics.com/#profiling

And attach it to this issue?

Memory profile for 1.63.0 version will be useful as well.

[pollynesterovich]

We tested upgrading to 1.65.0 and everything works fine, but problems with v1.66.0-1.66.2.

1.63.0.zip 1.66.0.zip

[pollynesterovich]

Please tell me when the answer will be given, the memory usage when upgrading to 1.66 or 1.67 versions has almost doubled, how can I fix this?