Elastic-apm-agent jdbcHelper seems to use a lot of memory

Elastic Observability APM
21

With your latest snapshot i can confirm that the SEVERE message no longer appears for those threads when the undeploy happens :slight_smile: (for this thread heh)

catalina.out:SEVERE: The web application [/v4##k000346] appears to have started a thread named [pool-11-thread-1] but has failed to stop it. This is very likely to create a memory leak.
catalina.out:SEVERE: The web application [/v4##k000346] appears to have started a thread named [I/O dispatcher 1] but has failed to stop it. This is very likely to create a memory leak.
catalina.out:SEVERE: The web application [/v4##k000346] appears to have started a thread named [I/O dispatcher 2] but has failed to stop it. This is very likely to create a memory leak.

these are the only ones now appearing, which doesnt include yours! Yay!

2 Likes
22

@Paul_Harris1 thanks for your extremely valuable input! You really made it easy for us to pinpoint and fix!

Just to verify- you get the SEVERE logging for pool-11-thread-1 (a basic Java Executors thread?) and I/O dispatcher 1/2 (Apache HTTP client threads?) when not using an agent as well?

23

That's actually a fair question, my assumption was they were unrelated, but I hadn't verified that.

I removed the agent etc, and I can confirm those messages are still present without the elastic-apm agent.

Ultimately I think i'll start moving away from hot deploys!

24

Thanks a lot for confirming!!
These threads may actually demonstrate how Tomcat's memory-leak-prevention mechanism does serve it purpose on hot deploys :slight_smile:
Note that 1.9.0 version of our agent was released with this fix included.
Again, thanks for your insightful inputs!

25

awesome :slight_smile: It all got shipped to the application servers as soon as i saw the release notification :slight_smile:

26

This topic was automatically closed 20 days after the last reply. New replies are no longer allowed.