Objects are not cleaned up

[orgads]

When running a load test, I notice that sockets and most objects remain and never get cleaned up.

I'm using a mock server that accepts websocket connections, and simulates cognitive services responses.

I start the sessions with recognizer.startContinuousRecognitionAsync(), and stop them after a while by calling recognizer.close().

Memory consumption increases constantly at a high rate (about 8-10MB/s when the session rate is 20/s).

Using Google Chrome inspector, I see that sockets and many other objects remain alive long after sessions are terminated.

[rhurey]

That's unexpected. I'll put an item into our work queue to dig in deeper.

What if you call stopContinuousRecognitionAsync() instead of close()?

[orgads]

Hi,

Sorry for the (very) late response.

We checked it now, and it doesn't seem to improve.

[orgads]

@rhurey?

[oscholz]

@orgads, thank you for your follow up. We've added this to our backlog, and will update this issue when we know more. For the time being, we will close this issue, but we'll continue to track this on our backlog. If there is any urgency to this, please reactivate, and tell us what this is blocking.
Once we resolve the backlog item, we will update this ticket to let you know what the resolution was.

[orgads]

Because of this issue we just can't use the official package, and we stick with an old implementation of ms-bing-speech-service with several patches.

We do want to use this package, but until this is solved, it will crash every few minutes on load runs, so obviously this blocks us.

[glharper]

@orgads Would you happen to still have the mock server with fake cog. service responses available to pass on so that I could take a look at this?

[orgads]

https://github.com/orgads/mock-server

[orgads]

@oscholz @glharper ^

[glharper]

@orgads Thank you for taking the time to write this up and help us out. We found and fixed a rather large memory leak, and the fix will be in our next release, 1.13. Much appreciated! :-)