NFSv3 client hangs against solaris9 server (2.4.23-25)
From: Auke Kok
Date: Wed Mar 24 2004 - 15:54:55 EST
past rtfm, stfw, switching all my linux nodes back to nfsver=2
explicitly, here's why and how:
I have a solaris9 NFS server exporting home directories and other stuff
to about 40 linux nodes. RH's 2.4.20-SMP kernel nicely uses nfsver=3 but
the stock 2.4.23 and 2.4.25 kernels compiled from source do not work
sufficiently, err correction, they don't work usably at all.
The problem is the nfs clients think the server is unavailable all the
time, after reverting to 'soft' mounts the machines are usable, but data
corruption will appear sooner or later anyway, so that's not it.
I know the server is okay since I've tried mounting the nfs filesystems
with the following options:
The nodes and the server are connected over GigaBit, there is only a
single 3com gbit switch in between them. Re-mounting to nfsvers=2
immediately resolves the issue, but like I said, only compiled kernels
have the issue.
BTW my test case is a simple ./configure && make in
/home/sofar/src/ccache-2.3, which is on the nfs-mount. on nfsvers=2 it
takes 20 seconds to complete, and on v3 about 3 hours timing out all the
time (possibly forever with hard mounts).
I'm willing to debug the crap out of this since it is driving me nuts,
if anyone wants to contact me please do so! Using linux for 8 years,
getting paid as a linux sysadmin, free-time linux distro maintainer...
and I can't get nfsvers=3 to work !!!! (help!help!)
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/