Anyone had any luck with either of these?
2.6.x as a server:
I'm using nfs-kernel-server v1.0.5-1 (debian).
Config Ops:
CONFIG_NFSD=m
CONFIG_NFSD_V3=y
# CONFIG_NFSD_V4 is not set
CONFIG_NFSD_TCP=y
CONFIG_LOCKD=m
CONFIG_LOCKD_V4=y << I didn't configure this, where'd it come from?
CONFIG_EXPORTFS=m
CONFIG_SUNRPC=m
This is in /etc/exports:
/ vegeta(rw,no_root_squash,async,nohide)
on the machine vegeta, I mount this machine (kingkai:/ /mnt), cd to /mnt, do
ls, and ls hangs. client uses kernel 2.4.20 with NFSv3 enabled and it was
mounted v3.
2.6.x as a client:
If the server is using 2.4.x with the kernel nfs server, all's fine except
for nohide. I have a 2.4.21 alpha machine exporting / and /raid-0 with the
nohide option. If I mount it to /mnt and cd /mnt/raid-0 there's nothing
there. On 2.4.20 client this works. (NFSv3 here)
If the server is using the userspace nfs server (I don't like exporting all
those mount points with nohide and this works for me on most machines that I
access occasionally. I assume userspace is NFSv2. 2.2beta47-12) Kernel
2.6.x will simply say the nfs server is not responding if I do an ls. If I
copy a file (not using tab completion), it seems to work.
client options:
CONFIG_NFS_FS=m
CONFIG_NFS_V3=y
# CONFIG_NFS_V4 is not set
Now on a positive note, my hp scanjet 5370c works with 2.6.0-test2 great
where as 2.4 it doesn't. the NFS is holding me back.
P.S.
Vegeta: kernel 2.4.20, userspace nfs, inspiron 8100 1ghz pIII 256mb ram
NIC: intel 10/100
Kingkai: Kernel 2.6.0-test1, kernel nfs, home build 700mhz pIII 256mb ram
NIC: 3com 3cr990srv 10/100 MainBoard: MSI MS-6163 v2
Alpha: Kernel 2.4.21, kernel nfs, AlphaServer 1000A 4/266 160mb ram
NIC: Dec Tulip 10/100
All machines use ext3 as the local filesystem.
--
Lab tests show that use of micro$oft causes cancer in lab animals
> > 2.6.x as a server:
> > I'm using nfs-kernel-server v1.0.5-1 (debian).
> > Config Ops:
> > CONFIG_NFSD=m
> > CONFIG_NFSD_V3=y
> > # CONFIG_NFSD_V4 is not set
> > CONFIG_NFSD_TCP=y
> > CONFIG_LOCKD=m
> > CONFIG_LOCKD_V4=y << I didn't configure this, where'd it come
> > from?
>
> If either NFS_V3 or NFSD_V3, then LOCKD_V4 is automatically added.
Ok, I see.
> > CONFIG_EXPORTFS=m
> > CONFIG_SUNRPC=m
> > This is in /etc/exports:
> > / vegeta(rw,no_root_squash,async,nohide)
>
> "nohide" is incorrect here.
previous experience. IIRC, if I didn't have it on /, I couldn't see /usr,
/raid-0, or others (2.4.x).
> > ls, and ls hangs. client uses kernel 2.4.20 with NFSv3 enabled and it was
> > mounted v3.
>
> I have a hunch what this might be, and it'll require fixing both the
> kernel and nfs-utils ;-(
>
> Could you try using
> mount -t nfsd nfsd /proc/fs/nfs
This worked. Of course I also removed nohide from / and added the rest of
my FSs. There were 4 in all on this test machine. / /usr /usr/src and
/home. All are exported as above except / has nohide removed.
> before starting the nfs-kernel-server (particularly before starting
> mountd or running exportfs).
>
> Also, if you are brave, could you try with the following two patches,
> one against linux-2.6.0-test2 and one against nfs-utils. Then run the
> nfs kernel server *without* mounting /proc/fs/nfs first.
I did not try the patches. This work around seems to be what I was looking
for. I'll wait for the patches to be merged. I did not patch the
nfs-kernel-server either.
I tried 2.6.x as a client to a v2 server running the userland server. I do
get a ton of nfs not responding messages, but I did a find /mnt/usr on it.
Doesn't seem to be going into D state this time.
--
Lab tests show that use of micro$oft causes cancer in lab animals