Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

BiglyBT wedges nearly daily #3217

Closed
runderwo opened this issue Mar 31, 2024 · 20 comments
Closed

BiglyBT wedges nearly daily #3217

runderwo opened this issue Mar 31, 2024 · 20 comments

Comments

@runderwo
Copy link

Hi, I just migrated from Azureus to BiglyBT 3.4.0 in Debian and while things are mostly working fine, the program completely hangs about once a day. Even kill -TERM is not responding and SIGKILL must be issued to get rid of the process. While Azureus was full of bugs, this was not one of them :-)

The JVM is 17.0.10+7-1~deb12u1 from Debian.

What should I do to best gather information that would be needed to find the root cause?

  • OS and version
  • BiglyBT Version Number
  • For *nix: Desktop Environment
@parg
Copy link
Contributor

parg commented Apr 1, 2024

@runderwo
Copy link
Author

runderwo commented Apr 1, 2024

Edit: nevermind, I forgot I need to manipulate a network namespace in my case. I'll check on it next time.

@runderwo
Copy link
Author

runderwo commented Apr 2, 2024

I found that curl connects but completely hangs with that query. It's as if the JVM is completely wedged. Only a kill -9 gets rid of the process.

@parg
Copy link
Contributor

parg commented Apr 2, 2024

Check the Debug_1/2.log files (and maybe SlowUI_1/2.log) in case anything is being logged prior to things hanging up

@runderwo
Copy link
Author

runderwo commented Apr 4, 2024

Ok, I think I got some useful logs from a hung instance.
debug.zip

@parg
Copy link
Contributor

parg commented Apr 4, 2024

Are you actually using I2P? There's a bunch of errors regarding failures contacting a router on "10.0.0.1:7654". If not then disable the I2P Helper plugin.

You have a large number of UDP listener threads running on port 0 which makes no sense...

Perhaps you could send over a full debug.zip (generate it from the Help menu) - email it to paul@biglybt.com if you want

@runderwo
Copy link
Author

runderwo commented Apr 4, 2024

I have not configured anything to listen on UDP port 0 - perhaps it's some artifact of migrating an old config from Vuze? The i2p configuration is correct. Are there other specific logs I could provide? debug.zip is far too detailed.

@parg
Copy link
Contributor

parg commented Apr 5, 2024

VPNHelper plugin installed?

You can generate "config changes" - go to Help->Advanced->Show Configuration Changes

You might want to sanitize it before posting it.

@runderwo
Copy link
Author

runderwo commented Apr 7, 2024

I don't even see a VPNHelper plugin in the available list, and definitely not installed.

Here is the output you asked for with parts marked as snipped:
configchanges.txt

@parg
Copy link
Contributor

parg commented Apr 7, 2024

Nothign sticks out there :( Do you really use SOCKS? It generally sucks.

@runderwo
Copy link
Author

runderwo commented Apr 7, 2024

Yes, to force all traffic with non-i2p destination through tor. I agree, it's not the greatest performance-wise but it's good enough in practice and shouldn't cause the whole process to get stuck. Is there any problem you see with using the Java 17 JRE from Debian?

@runderwo
Copy link
Author

Seems like it gets very slow before wedging which makes me think it might be a GC death spiral. Do you have a set of JVM flags and/or process for tracking down leaks?

@parg
Copy link
Contributor

parg commented Apr 17, 2024

@runderwo
Copy link
Author

runderwo commented Apr 19, 2024

There appears to be a memory leak. No matter how high I set -Xmx, eventually "alloc" bumps up against it, "free" dwindles to a few megabytes and the program becomes unresponsive. I have noticed in the log that "alloc" generally merely creeps up slowly over a period of hours; at times, it jumps more than a gigabyte in a matter of seconds, but I guess that could be the JVM allocating additional heap in big chunks. Is there a process for obtaining and analyzing a heap dump or do I need to figure it out myself?

@parg
Copy link
Contributor

parg commented Apr 19, 2024

You could try https://visualvm.github.io/

@runderwo
Copy link
Author

Heap objects in screenshot
heap

@parg
Copy link
Contributor

parg commented Apr 22, 2024

How many active downloads (i.e. Downloading or Seeding) do you have? 11,000 ?

@runderwo
Copy link
Author

I noticed that max active torrents was set to 0, so I set it to 100. Maybe this was an artifact of migrating. Let's see if that behaves better.

@runderwo
Copy link
Author

Looks like that configuration error was the problem since memory usage is much more consistent now. Thanks for your help! Perhaps the setting did not get migrated correctly, in case that's worth looking into.

@parg
Copy link
Contributor

parg commented Apr 26, 2024

Glad it worked out!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants