Possible memory leak

[SimonBradley1993]

Hi,

We have Vroom docker deployed on AWS and are hitting it with a steady, but quite high, number of requests. We're seeing the memory usage of the container that Vroom docker is in consistently creeping up. Upon restart of this service the memory usage drops drastically, but then begins to creep up again.

I'm not sure what would be causing this, we wrap the Vroom docker image in our own Dockerfile to be able to inject config, but otherwise it's the plain Vroom docker image.

We're running the latest version, 1.14.0.

Any help would be great.

Thanks!

[jcoupey]

Hard to tell here if the problem lies in VROOM or the container context. The best way to move forward on this would be to run vroom through valgrind (either inside or outside a container) with a batch of instances you have that trigger the problem.

[SimonBradley1993]

I've run some tests using leaks on my mac (valgrind wouldn't work for my OS version). I haven't run the tests on the docker container, but I ran them against Vroom itself, and vroom-express.

I found that Vroom doesn't have any memory leaks, directly or within vroom-express, but the vroom-express process will continually use more memory and not release it again. This was done locally by hitting vroom-express with a few medium sized requests and I managed to get the memory usage up to 30Mb and saw it not drop after some time.

This would match what we've seen in production, except production is handling hundreds of req/s.

Should I move this over to the vroom-express repo?

[jcoupey]

Thanks for reporting! This is more suited for the vroom-express repo indeed then.