* NFS rpc and stale handles on 2.6.x servers @ 2004-01-26 1:03 hanasaki 2004-01-30 0:46 ` [NFS] " Neil Brown 0 siblings, 1 reply; 6+ messages in thread From: hanasaki @ 2004-01-26 1:03 UTC (permalink / raw) To: nfs The below is being reported, on and off, when hitting nfs-kernel-servers running on 2.6.0 and 2.6.1 Could someone tell me if this is smoe bug or what? Thanks RPC request reserved 0 but used 124 Debian sarge nfs-kernel-server am-untils nfsv3 over tcp ------------------------------------------------------- The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA. http://www.eclipsecon.org/osdn _______________________________________________ NFS maillist - NFS@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [NFS] NFS rpc and stale handles on 2.6.x servers 2004-01-26 1:03 NFS rpc and stale handles on 2.6.x servers hanasaki @ 2004-01-30 0:46 ` Neil Brown 2004-01-30 1:25 ` Mike Fedyk 2004-01-30 11:55 ` Roman Kagan 0 siblings, 2 replies; 6+ messages in thread From: Neil Brown @ 2004-01-30 0:46 UTC (permalink / raw) To: hanasaki; +Cc: nfs, linux-kernel On Sunday January 25, hanasaki@hanaden.com wrote: > The below is being reported, on and off, when hitting nfs-kernel-servers > running on 2.6.0 and 2.6.1 Could someone tell me if this is smoe bug or > what? Thanks > RPC request reserved 0 but used 124 > > Debian sarge > nfs-kernel-server > am-untils > nfsv3 over tcp > stale file handles is a known bug that is fixed in the but BK and will be in 2.6.3. The "RPC request reserved 0 ..." is very odd. It does immediately indicate a major problem, but it should be fixed, if only I could figure out what was causing it. I might put come more info into the message so future bug reports will tell me more. Thanks, NeilBrown ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [NFS] NFS rpc and stale handles on 2.6.x servers 2004-01-30 0:46 ` [NFS] " Neil Brown @ 2004-01-30 1:25 ` Mike Fedyk 2004-01-30 1:36 ` Neil Brown 2004-01-30 11:55 ` Roman Kagan 1 sibling, 1 reply; 6+ messages in thread From: Mike Fedyk @ 2004-01-30 1:25 UTC (permalink / raw) To: Neil Brown; +Cc: hanasaki, nfs, linux-kernel On Fri, Jan 30, 2004 at 11:46:31AM +1100, Neil Brown wrote: > On Sunday January 25, hanasaki@hanaden.com wrote: > > The below is being reported, on and off, when hitting nfs-kernel-servers > > running on 2.6.0 and 2.6.1 Could someone tell me if this is smoe bug or > > what? Thanks > > RPC request reserved 0 but used 124 > > > > Debian sarge > > nfs-kernel-server > > am-untils > > nfsv3 over tcp > > > > stale file handles is a known bug that is fixed in the but BK and will > be in 2.6.3. do you mean 2.6.2? I've merged the nfsd stale file handles into 2.6.1-bk2 and it is working fine on a nfs server here... ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [NFS] NFS rpc and stale handles on 2.6.x servers 2004-01-30 1:25 ` Mike Fedyk @ 2004-01-30 1:36 ` Neil Brown 2004-01-30 2:11 ` hanasaki 0 siblings, 1 reply; 6+ messages in thread From: Neil Brown @ 2004-01-30 1:36 UTC (permalink / raw) To: Mike Fedyk; +Cc: hanasaki, nfs, linux-kernel On Thursday January 29, mfedyk@matchmail.com wrote: > On Fri, Jan 30, 2004 at 11:46:31AM +1100, Neil Brown wrote: > > On Sunday January 25, hanasaki@hanaden.com wrote: > > > The below is being reported, on and off, when hitting nfs-kernel-servers > > > running on 2.6.0 and 2.6.1 Could someone tell me if this is smoe bug or > > > what? Thanks > > > RPC request reserved 0 but used 124 > > > > > > Debian sarge > > > nfs-kernel-server > > > am-untils > > > nfsv3 over tcp > > > > > > > stale file handles is a known bug that is fixed in the but BK and will > > be in 2.6.3. > > do you mean 2.6.2? Yeh, 2.6.2 as well.. But definitely 2.6.3 :-) > > I've merged the nfsd stale file handles into 2.6.1-bk2 and it is working > fine on a nfs server here... good, thanks. NeilBrown ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [NFS] NFS rpc and stale handles on 2.6.x servers 2004-01-30 1:36 ` Neil Brown @ 2004-01-30 2:11 ` hanasaki 0 siblings, 0 replies; 6+ messages in thread From: hanasaki @ 2004-01-30 2:11 UTC (permalink / raw) To: Neil Brown; +Cc: Mike Fedyk, nfs, linux-kernel Add this to your exportfs "no_subtree_check" It seems to be a temp workaround. Searching the web, the only issue with this option seems to be a minor performance hit (big issue for large systems). Any chance of getting the patch 2.6.2? www.kernel.org looks like its still on an RC2 of this. Neil Brown wrote: > On Thursday January 29, mfedyk@matchmail.com wrote: > >>On Fri, Jan 30, 2004 at 11:46:31AM +1100, Neil Brown wrote: >> >>>On Sunday January 25, hanasaki@hanaden.com wrote: >>> >>>>The below is being reported, on and off, when hitting nfs-kernel-servers >>>>running on 2.6.0 and 2.6.1 Could someone tell me if this is smoe bug or >>>>what? Thanks >>>> RPC request reserved 0 but used 124 >>>> >>>>Debian sarge >>>>nfs-kernel-server >>>>am-untils >>>>nfsv3 over tcp >>>> >>> >>>stale file handles is a known bug that is fixed in the but BK and will >>>be in 2.6.3. >> >>do you mean 2.6.2? > > > Yeh, 2.6.2 as well.. But definitely 2.6.3 :-) > > >>I've merged the nfsd stale file handles into 2.6.1-bk2 and it is working >>fine on a nfs server here... > > > good, thanks. > NeilBrown ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [NFS] NFS rpc and stale handles on 2.6.x servers 2004-01-30 0:46 ` [NFS] " Neil Brown 2004-01-30 1:25 ` Mike Fedyk @ 2004-01-30 11:55 ` Roman Kagan 1 sibling, 0 replies; 6+ messages in thread From: Roman Kagan @ 2004-01-30 11:55 UTC (permalink / raw) To: Neil Brown; +Cc: linux-kernel On Fri, Jan 30, 2004 at 10:16:05AM +0000, Neil Brown wrote: > The "RPC request reserved 0 ..." is very odd. It does immediately > indicate a major problem, but it should be fixed, if only I could > figure out what was causing it. In case it helps: having enabled svcsock debugging by echo $[0x0100] > /proc/sys/sunrpc/rpc_debug I've noticed that those messages always appear in the same pattern: svc: server c7a3d200 waiting for data (to = 3600000) svc: socket c6dbfb00(inet c769f220), write_space busy=0 svc: socket c2a9fac0 TCP data ready (svsk c6dbf980) svc: socket c2a9fac0 served by daemon c7a3d200 svc: socket c2a9fac0 TCP data ready (svsk c6dbf980) svc: socket c2a9fac0 busy, not enqueued svc: server c7a3d200, socket c6dbf980, inuse=1 svc: tcp_recv c6dbf980 data 1 conn 0 close 0 svc: socket c6dbf980 recvfrom(c6dbf9d8, 0) = 4 svc: TCP record, 2584 bytes svc: socket c6dbf980 recvfrom(c6d06a18, 1512) = 2584 svc: TCP complete record (2584 bytes) svc: socket c2a9fac0 served by daemon c6efe000 svc: got len=2584 svc: socket c2a9fac0 busy, not enqueued svc: socket c6dbf980 sendto([c436b000 140... ], 140) = 140 (addr 43e17cc1) svc: socket c2a9fac0 busy, not enqueued svc: server c7a3d200 waiting for data (to = 3600000) svc: server c7a3d200, socket c25007a0, inuse=1 svc: tcp_recv c25007a0 data 0 conn 0 close 1 svc: svc_delete_socket(c25007a0) svc: server socket destroy delayed svc: got len=0 RPC request reserved 0 but used 140 svc: releasing dead socket svc: server c7a3d200 waiting for data (to = 3600000) Note that "tcp_recv" with this set of parameters (data=0 conn=0 close=1) is always correlated with "RPC request reserved ...", and also the "used" request length matches the message length in "sendto" on the seemingly unrelated socket. Unfortunately I don't understand the code well enough to make a better bug report, but feel free to ask me to test your patches if you can't reproduce the problem in your setup. Roman. ^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2004-01-30 11:55 UTC | newest] Thread overview: 6+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2004-01-26 1:03 NFS rpc and stale handles on 2.6.x servers hanasaki 2004-01-30 0:46 ` [NFS] " Neil Brown 2004-01-30 1:25 ` Mike Fedyk 2004-01-30 1:36 ` Neil Brown 2004-01-30 2:11 ` hanasaki 2004-01-30 11:55 ` Roman Kagan
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.