Subject: Re: NFS server hangs under 1.4.1
To: None <port-i386@NetBSD.ORG>
From: Kent Polk <kent@tiamat.goathill.org>
List: port-i386
Date: 03/06/2000 17:12:19
On 5 Mar 2000 22:40:01 -0600, Steve wrote:
>I've had similar NFS grief under 1.4.1. Mostly on very
>good hardware (P2-450/128MB)
>
>Sometimes the machines hang and sometimes a
>full system reboot will happen randomly under load.
>It's next to impossible to debug/trace. It usually
>happens at least once every 2 days.
>
>I do think I might have isolated it to the ep device
>driver/3C509 NIC cards as I had two servers that were
>exhibiting the behavior. Both servers are dual-homed.
I have had an NFS problem with ep/3C509 as a NFS client for almost
a year. The server is a Solaris Ultra 5 and this problem only exists
with the NetBSD (1.4.0/1.4.1) NFS client. The problem is that when
*sending* files around 2MB and larger, the NetBSD client starts
hanging for typically from about 30 seconds to a minute or so and
sometimes it simply fails. When it fails, nfsstat indicates the
failure, but otherwise I don't really see anything in the logs that
indicates a problem that I understand. When the nfs client hangs,
the NetBSD box almost comes to a halt. Note that the NFS *server*
does not hang, only the client. I can perform NFS transfers with
no apparent problem from another Solaris client while the NetBSD
client is hung.
Exactly what should I be looking for to examine this problem?
Thanks
-----------------
nfsstat - Client Info:
Rpc Counts:
Getattr Setattr Lookup Readlink Read Write Create Remove
33 0 115 0 0 9697 4 3
Rename Link Symlink Mkdir Rmdir Readdir RdirPlus Access
0 0 0 0 0 6 0 33
Mknod Fsstat Fsinfo PathConf Commit GLease Vacate Evict
0 20 1 0 948 0 0 0
Rpc Info:
TimedOut Invalid X Replies Retries Requests
0 0 2744 2766 10860
Cache Info:
Attr Hits Misses Lkup Hits Misses BioR Hits Misses BioW Hits Misses
139 33 31 115 0 0 -3958 9697
BioRLHits Misses BioD Hits Misses DirE Hits Misses
0 0 22 6 18 9