On Wed, Mar 08, 2006 at 12:30:02AM +0000, Miguel Lopes Santos Ramos wrote: > > From: Kris Kennaway <[EMAIL PROTECTED]> > > Subject: Re: rpc.lockd brokenness (2) > > > [...] > > but there's no evidence in the trace that it ever tries to write. Can > > you also obtain a ktrace -i dump from cron? > > The file remains empty. > I really don't know enough about NFS, but isn't that getattr message repeated > some seconds latter, and repeated... (even though it always gets an answer)
They have different file handles (which weren't identified in the previous trace, i.e. they predate the start of the trace), so it could just be background noise from other reads on the system. > The ktrace is in http://mega.ist.utl.pt/~mlsr/ktrace.txt > > I'm not sure it's good. I can't see cron.pid there. > I had to reboot to end the process, otherwise I couldn't kill cron and > the trace didn't grow either. I wonder if something else is going wrong and it's not rpc.lockd at all. > > Also while you're there, could you obtain a binary format tcpdump > > (tcpdump -w) instead? This may be parsed with tools like ethereal > > which will help with the analysis. > > The tcpdump -w is in http://mega.ist.utl.pt/~mlsr/nfs.bin It looks like this wasn't made using -s 0 - sorry if I wasn't explicit. Kris
pgpqJAnNuUvYE.pgp
Description: PGP signature
