On Saturday 22 August 2009 05:56:09 Dennis Nezic wrote: > It appears that the problem only occurs in conjunction with FMS (0.3.41 > at least). That is, freenet (I believe the problem started with 1229, > and it's still here with 1232. I really should test 1228) will run > normally -- and only after I launch FMS does it skyrocket to consume all > available cpu power. Closing FMS doesn't alleviate the problem. > Noticeable differences in "Thread usage": FCP input handler (1) and FCP > output handler (1) and possibly FailureTable offers executor (1) appear > after FMS starts, and REMAIN even after I close FMS. My only "fix" is to > restart the node :|. > > Smells like some kind of infinite loop in the FCP handlers?
If so, a thread dump ought to demonstrate it. Please post one. > > On Mon, 17 Aug 2009 18:47:27 -0400, Dennis Nezic wrote: > > Ever since updating to build 1229, my java-freenet process takes up > > far too much cpu time than it should -- nothing is in the global > > queue and I deleted my node.db4o. Advanced stats page shows 1 thread > > is running with priority 1, 26 with prio 5 (19 waiting), 57 with prio > > 7 (16 waiting). But which is that 1 culprit thread?: > > > > RequestSender: 36 (29.5%) > > RequestStarter$SenderThread: 17 (13.9%) > > Pooled thread awaiting work : 15 (12.3%) > > Network Interface Acceptor: 5 (4.1%) > > Finish CHK transfer: 4 (3.3%) > > BlockTransmitter: 3 (2.5%) > > BlockTransmitter:sendAsync: 2 (1.6%) > > FCP input handler: 2 (1.6%) > > FCP output handler: 2 (1.6%) > > UdpSocketHandler: 2 (1.6%) > > Announcement sender: 1 (0.8%) > > Background block encoder: 1 (0.8%) > > CHK Insert starter : 1 (0.8%) > > CHK Request starter : 1 (0.8%) > > Client database access thread: 1 (0.8%) > > Compression scheduler: 1 (0.8%) > > DNSRequester thread: 1 (0.8%) > > Datastore checker: 1 (0.8%) > > DestroyJavaVM: 1 (0.8%) > > Diffie-Hellman-Precalc: 1 (0.8%) > > FCP server: 1 (0.8%) > > FailureTable offers executor: 1 (0.8%) > > Finalizer: 1 (0.8%) > > HTTP socket handler: 1 (0.8%) > > IP address re-detector: 1 (0.8%) > > PacketSender thread: 1 (0.8%) > > Plug: 1 (0.8%) > > Reference Handler: 1 (0.8%) > > RequestHandler: 1 (0.8%) > > SSK Insert starter : 1 (0.8%) > > SSK Request starter : 1 (0.8%) > > Send throttled SSK data: 1 (0.8%) > > Signal Dispatcher: 1 (0.8%) > > SimpleToadletServer: 1 (0.8%) > > Store-CHK-cache-Cleaner: 1 (0.8%) > > Store-CHK-store-Cleaner: 1 (0.8%) > > Store-PUBKEY-cache-Cleaner: 1 (0.8%) > > Store-PUBKEY-store-Cleaner: 1 (0.8%) > > Store-SSK-cache-Cleaner: 1 (0.8%) > > Store-SSK-store-Cleaner: 1 (0.8%) > > SwapRequestSender: 1 (0.8%) > > Text mode client interface: 1 (0.8%) > > Wrapper-Control-Event-Monitor: 1 (0.8%) > > db4o WeakReference collector: 1 (0.8%) > > > > My total cpu usage remains at 100% for hours -- in other words, build > > 1229 is effectively useless here :\.
signature.asc
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
