> > I tracked down the cause of my mysterious Sun RPC / NIS failures. These
> > appeared with the 2.1.127-pre3 patches, and have been relentless. Turns
> > out that there's a conflict between one of HJ's kernel patches (supplied
> > with knfsd-1022) and the recent scheduler changes. I've not yet tracked
> > down exactly which one, but suspect the mods to sunrpc/sched.c and
> > sched.h.
> >
> > All I can verify at the moment is that 2.1.127-pre7 without knfsd has
> > an operational RPC layer.
>
> Can you apply all my relevant knfsd patches and then try
>
> # cd .../linux
> # patch -P2 -R < my_patch_here
>
(snip)
Yes, your sched changes were the problem. With everything else but
them applied, it boots and seems to work. One bit of
remaining strangeness:
On my SMP (Dual PPro) box, the first use of RPC by any client (local or
remote) hangs for quite a while before timing out. Subsequently, all
other RPC calls work:
On the host:
rpc.kmountd and knfsd each pause for quite a while at boot. When they
come up, the server is functional and can be mounted by clients.
next (still on the host):
$ kshowmount -e
(hangs for quite a bit)
...Message about RPC timeout..
$ kshowmount -e
..works normally
On client:
$ kshowmount -e server
..hangs and timesout
$ kshowmount -e server
..works
(all other clients then work, too)
It's almost as if something plugs every service until it is forcibly timed
out and reinitialized (if that makes any sense).
Again, I'm only seeing this on the SMP box. I do not run SMP kernels on
the non-SMP boxes, FWIW. All kernels are built under gcc 2.7.2.3 (none of
the last few egcs snapshots will compile for me, and release 1.1b fails
miserably on 2.1.127 and 128).
Regards,
Steve
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/