-
Notifications
You must be signed in to change notification settings - Fork 217
Exchange doesn't disconnect gracefully #35
Comments
@jpink Yeah, I can reproduce that and it's bug of exchanges based on |
Did anyone ever do anything about this? It's the nioEventLoopGroup thread that's still running I think, but I don't know enough about these things. I got around it by assigning the result of |
This one is pretty annoying, sure I can add a System.exit() call so that I'm not chasing down PIDs but it breaks WAR hot deploy as I need to stop tomcat and restart it to pick up code changes, I should be able to just trigger the deploy from rebuilding. |
This issue also seems to effect Gemini. |
Possibly related - I get an NPE whenever I disconnect from GDAX due to a background socket thread:
|
I'm trying to make some progress with all this stuff on #191 (a little bit blindly as I'm not familiar with Netty) and am keeping the PR rebased against develop. In the meantime, in the absence of PR approvals right now, I am producing snapshot builds myself which pull together
Just make sure you're running your Maven build with |
Test to simulate thread leak:
|
Try to update netty service dependency |
looks like i have fixed problem here: #225 |
Not sure if related but I just go my process dying with Too many open files, and it turns out Netty spun up a LOT of threads, at the moment 651 alive of format:
and clearly a lot more were actually created, about 4 millions judging by the thread names above, all with the same stacktrace:
and 4150 (!) sockets opened... |
When I repeat following steps as in example:
But program doesn't stop! There is still threads running somewhere.
I'm using Spring Boot web app and I'm stopping web server via POST /shutdown request. All other services are then shutdown.
This is not big deal. Currently I call System.exit() 15 s after that request. But this is annoying and smells that there is thread leak somewhere.
The text was updated successfully, but these errors were encountered: