* nfs/nfsd gets stuck on a write
@ 2003-08-06 10:46 Wichert Akkerman
2003-08-06 11:02 ` bert hubert
0 siblings, 1 reply; 3+ messages in thread
From: Wichert Akkerman @ 2003-08-06 10:46 UTC (permalink / raw)
To: linux-kernel; +Cc: nfs
I seem to have hit a nfs problem between two 2.6.0-test2 machines.
Mounting works properly, but at some point when logging in with /home
over NFS things get stuck. A tcpdump reveals the following:
12:37:38.601967 192.168.1.50.3218252557 > 192.168.1.2.nfs: 132 getattr fh Unknown/1 (DF)
12:37:38.617871 192.168.1.2.nfs > 192.168.1.50.3218252557: reply ok 112 (DF)
12:37:38.618274 192.168.1.50.3235029773 > 192.168.1.2.nfs: 132 getattr fh Unknown/1 (DF)
12:37:38.627868 192.168.1.2.nfs > 192.168.1.50.3235029773: reply ok 112 (DF)
12:37:38.628238 192.168.1.50.3251806989 > 192.168.1.2.nfs: 132 getattr fh Unknown/1 (DF)
12:37:38.637865 192.168.1.2.nfs > 192.168.1.50.3251806989: reply ok 112 (DF)
12:37:38.638276 192.168.1.50.3268584205 > 192.168.1.2.nfs: 132 getattr fh Unknown/1 (DF)
12:37:38.646863 192.168.1.2.nfs > 192.168.1.50.3268584205: reply ok 112 (DF)
12:37:38.647348 192.168.1.50.3285361421 > 192.168.1.2.nfs: 132 getattr fh Unknown/1 (DF)
12:37:38.655862 192.168.1.2.nfs > 192.168.1.50.3285361421: reply ok 112 (DF)
12:37:38.656294 192.168.1.50.3302138637 > 192.168.1.2.nfs: 140 lookup fh Unknown/1 "backgrounds" (DF)
12:37:38.664860 192.168.1.2.nfs > 192.168.1.50.3302138637: reply ok 228 (DF)
12:37:38.665190 192.168.1.50.3318915853 > 192.168.1.2.nfs: 132 getattr fh Unknown/1 (DF)
12:37:38.685857 192.168.1.2.nfs > 192.168.1.50.3318915853: reply ok 112 (DF)
12:37:38.687035 192.168.1.50.3335693069 > 192.168.1.2.nfs: 1472 write fh Unknown/1 6189 bytes @ 0x000000000 (frag 41128:1480@0+)
12:37:38.687155 192.168.1.50 > 192.168.1.2: udp (frag 41128:1480@1480+)
12:37:38.687278 192.168.1.50 > 192.168.1.2: udp (frag 41128:1480@2960+)
12:37:38.687401 192.168.1.50 > 192.168.1.2: udp (frag 41128:1480@4440+)
12:37:38.687429 192.168.1.50 > 192.168.1.2: udp (frag 41128:432@5920)
12:37:39.044132 192.168.1.50.3335693069 > 192.168.1.2.nfs: 1472 write fh Unknown/1 6189 bytes @ 0x000000000 (frag 41129:1480@0+)
12:37:39.044250 192.168.1.50 > 192.168.1.2: udp (frag 41129:1480@1480+)
12:37:39.044371 192.168.1.50 > 192.168.1.2: udp (frag 41129:1480@2960+)
12:37:39.044496 192.168.1.50 > 192.168.1.2: udp (frag 41129:1480@4440+)
12:37:39.044523 192.168.1.50 > 192.168.1.2: udp (frag 41129:432@5920)
12:37:39.758006 192.168.1.50.3335693069 > 192.168.1.2.nfs: 1472 write fh Unknown/1 6189 bytes @ 0x000000000 (frag 41130:1480@0+)
12:37:39.758125 192.168.1.50 > 192.168.1.2: udp (frag 41130:1480@1480+)
12:37:39.758248 192.168.1.50 > 192.168.1.2: udp (frag 41130:1480@2960+)
12:37:39.758371 192.168.1.50 > 192.168.1.2: udp (frag 41130:1480@4440+)
12:37:39.758399 192.168.1.50 > 192.168.1.2: udp (frag 41130:432@5920)
12:37:41.185748 192.168.1.50.3335693069 > 192.168.1.2.nfs: 1472 write fh Unknown/1 6189 bytes @ 0x000000000 (frag 41131:1480@0+)
12:37:41.185866 192.168.1.50 > 192.168.1.2: udp (frag 41131:1480@1480+)
12:37:41.185988 192.168.1.50 > 192.168.1.2: udp (frag 41131:1480@2960+)
12:37:41.186110 192.168.1.50 > 192.168.1.2: udp (frag 41131:1480@4440+)
12:37:41.186138 192.168.1.50 > 192.168.1.2: udp (frag 41131:432@5920)
Since 192.168.1.2 does not respond 192.168.1.50 keeps sending that data
and basically gets stuck waiting for an answer. A complete tcpdump from
the booting of the client to it getting stuck can be found at
http://home.wiggy.net/lkml/nfs.tcpdump .
Both machines have a similar configuration: 2.6.0-test2 with NFS and NFS
over TCP enabled. There are no firewalls blocking traffic between those
two machines.
Wichert.
--
Wichert Akkerman <wichert@wiggy.net> It is simple to make things.
http://www.wiggy.net/ It is hard to make things simple.
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: nfs/nfsd gets stuck on a write
2003-08-06 10:46 nfs/nfsd gets stuck on a write Wichert Akkerman
@ 2003-08-06 11:02 ` bert hubert
2003-08-06 11:16 ` Wichert Akkerman
0 siblings, 1 reply; 3+ messages in thread
From: bert hubert @ 2003-08-06 11:02 UTC (permalink / raw)
To: linux-kernel, nfs
On Wed, Aug 06, 2003 at 12:46:13PM +0200, Wichert Akkerman wrote:
> I seem to have hit a nfs problem between two 2.6.0-test2 machines.
> Mounting works properly, but at some point when logging in with /home
> over NFS things get stuck. A tcpdump reveals the following:
I've had problems like that go away by explicitly mounting -o tcp.
> Both machines have a similar configuration: 2.6.0-test2 with NFS and NFS
> over TCP enabled. There are no firewalls blocking traffic between those
> two machines.
Your dump shows UDP, so give it a try.
--
http://www.PowerDNS.com Open source, database driven DNS Software
http://lartc.org Linux Advanced Routing & Traffic Control HOWTO
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: nfs/nfsd gets stuck on a write
2003-08-06 11:02 ` bert hubert
@ 2003-08-06 11:16 ` Wichert Akkerman
0 siblings, 0 replies; 3+ messages in thread
From: Wichert Akkerman @ 2003-08-06 11:16 UTC (permalink / raw)
To: linux-kernel, nfs; +Cc: bert hubert
Previously bert hubert wrote:
> I've had problems like that go away by explicitly mounting -o tcp.
Mounting with -o tcp does indeed seem to be a working work-around. It
does feel a lot more sluggish/high latency than udp though.
Wichert.
--
Wichert Akkerman <wichert@wiggy.net> It is simple to make things.
http://www.wiggy.net/ It is hard to make things simple.
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2003-08-06 11:16 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-08-06 10:46 nfs/nfsd gets stuck on a write Wichert Akkerman
2003-08-06 11:02 ` bert hubert
2003-08-06 11:16 ` Wichert Akkerman
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).