On Wednesday 21 October 2009 23:19:18 Dennis Nezic wrote: > On Wed, 21 Oct 2009 15:50:01 -0500, CyberLeo wrote: > > About once every 1-2 weeks, my node will shut down completely. > > > > In the wrapper log, I see: > > INFO | jvm 2 | 2009/10/20 16:30:59 | Out of memory: Emergency > > shutdown to protect database integrity in progress... > > > > followed by a full thread dump and db shutdown, and then > > > > ERROR | wrapper | 2009/10/20 16:33:09 | Shutdown failed: Timed out > > waiting for the JVM to terminate. > > ERROR | wrapper | 2009/10/20 16:33:09 | JVM did not exit on request, > > terminated > > INFO | wrapper | 2009/10/20 16:33:09 | JVM exited on its own while > > waiting to kill the application. > > STATUS | wrapper | 2009/10/20 16:33:09 | JVM exited in response to > > signal SIGKILL (9). > > STATUS | wrapper | 2009/10/20 16:33:09 | <-- Wrapper Stopped > > > > then terminates without restarting. Is a total shutdown instead of a > > simple restart intended? The machine itself runs comfortably, and > > never oomkills, and the wrapper has wrapper.java.maxmemory=256. > > > > I can upload the entire wrapper.log and wrapper.conf someplace, if > > needed. > > > > Thanks! > > I have this problem too. Have had it for a long time. But at least the > recent releases shut down gracefully. What I still don't understand is > if it's stable the first week (and it is), why can't the internal > garbage-men clean things up to roughly that state? Perhaps there should > be better internal control of memory allocation to prevent the node > from using up all the allocated memory -- have it use up the last 5% > (or whatever) of memory only for critical situations.
That's what's called a memory leak. If every day we leak 10MB of stuff, which we can't clean up because it's a leak - something still points to it - then by the end of a couple of weeks we're stuffed. Hopefully the next build (try it now if you can build from git) will improve this slightly. Has it improved since October? I agree the fact that the wrapper doesn't restart the node is strange. There is a bug filed for that.
Description: This is a digitally signed message part.
_______________________________________________ Support mailing list Support@freenetproject.org http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:support-requ...@freenetproject.org?subject=unsubscribe