nfs probs with fc2

[decals74]

hello all. i'm having probs mounting nfs filesystems with an fc2 nfs-server and fc2 client. error msg is always 'Permission denied'.

here's what i've tried in /etc/exports [3 different attempted configs exported via `exportfs -vr`]:

# attempt1
/zz bluto(rw,no_root_squash,sync)

# attempt2
/zz 192.168.0.222(rw,no_root_squash,sync)

#attempt3
/zz 192.168.0.0/24(rw,no_root_squash,sync)

/etc/sysconfig/nfs on the server is null and iptables is accepting on all chains.

/etc/hosts on the server contains:
127.0.0.1 localhost.localdomain localhost mongo
192.168.0.222 bluto

when i try to mount i get:

[root@bluto bin]# mount -t nfs -o rsize=8192,wsize=8192 mongo:/zz /home/steve/zz
mount: mongo:/zz failed, reason given by server: Permission denied

i've tried mounting with nfsvers=2 and nfsvers=3 options, no dice. i see this in the server's logs everytime it fails:

Jun 1 01:51:28 localhost rpc.mountd: authenticated mount request from bluto:783
for /zz (/zz)

here's an rpcinfo dump:

[root@mongo etc]# rpcinfo -p
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 32768 status
100024 1 tcp 32768 status
391002 2 tcp 32769 sgi_fam
100011 1 udp 768 rquotad
100011 2 udp 768 rquotad
100011 1 tcp 771 rquotad
100011 2 tcp 771 rquotad
100003 2 udp 2049 nfs
100003 3 udp 2049 nfs
100003 4 udp 2049 nfs
100003 2 tcp 2049 nfs
100003 3 tcp 2049 nfs
100003 4 tcp 2049 nfs
100021 1 udp 32785 nlockmgr
100021 3 udp 32785 nlockmgr
100021 4 udp 32785 nlockmgr
100021 1 tcp 32892 nlockmgr
100021 3 tcp 32892 nlockmgr
100021 4 tcp 32892 nlockmgr
100005 1 udp 784 mountd
100005 1 tcp 787 mountd
100005 2 udp 784 mountd
100005 2 tcp 787 mountd
100005 3 udp 784 mountd
100005 3 tcp 787 mountd

[root@mongo etc]# uname -a
Linux mongo 2.6.5-1.358 #1 Sat May 8 09:04:50 EDT 2004 i686 athlon i386 GNU/Linux

any suggestions?

[crackers]

Two dumb questions:

1) Can you "ping" mongo from bluto? (Is it reachable?)

2) Did you restart the NFS daemon after configuring /etc/exports?

I said they were dumb - and they've tripped me up more than once...

[decals74]

sup crackers.

yeah i can ping it from bluto and access ftp and apache on mongo as well.

i've been restarting nfs after making config changes via `/etc/init.d/nfs restart` and `service nfs restart` after getting suspicious feelings bout `exportfs -vr`.

[crackers]

Hey, I had to ask - it's hard to tell how much experience a poster has. Or what silly little bit they've overlooked. Not like I've ever done that last bit. Nope. Never.

Yeah, right... :D

[sniperX]

1. Please check the NFS directory permission.

2. The firewall setup for NFS.

3. The host allow in NFS setting.

[decals74]

sup sniper.

- the shared dir is 777
- iptables is down with ACCEPT policy on all chains
- /etc/hosts.allow and /etc/hosts.deny are wide open [null]

tcpdump while attempting mounts doesnt reveal anything interesting

i wonder if this is linked to the problem:

# /etc/init.d/nfs restart causes this to appear in /var/log/messages:

Jun 2 17:06:56 localhost rpc.mountd: Caught signal 15, un-registering and exiti
ng.
Jun 2 17:06:56 localhost nfs: rpc.mountd shutdown succeeded
Jun 2 17:07:00 localhost kernel: nfsd: last server has exited
Jun 2 17:07:00 localhost kernel: nfsd: unexporting all filesystems
Jun 2 17:07:00 localhost kernel: rpciod: active tasks at shutdown?!
Jun 2 17:07:00 localhost kernel: RPC: failed to contact portmap (errno -5).
Jun 2 17:07:00 localhost nfs: nfsd shutdown succeeded
Jun 2 17:07:00 localhost nfs: rpc.rquotad shutdown succeeded
Jun 2 17:07:00 localhost nfs: Shutting down NFS services: succeeded
Jun 2 17:07:00 localhost nfs: Starting NFS services: succeeded
Jun 2 17:07:00 localhost nfs: rpc.rquotad startup succeeded
Jun 2 17:07:00 localhost nfs: rpc.nfsd startup succeeded
Jun 2 17:07:20 localhost nfs: rpc.mountd startup succeeded

i've tried nfs2, nfs3... no dice. wtf?? i've resorted to using samba for now. am i the only person having nfs issues with fc2?? weird. thanks for the advice anyways.

[coyote4til7]

Following the same steps that work in RedHat 9, I've had no luck. Lots of fiddling based on things I've found here and there has gotten me no closer. I'm trying to use one FC2 (final) machine as an install server. I'm using /export for things I want to share via nfs.

I created /export/fedora-core-2
Within /export/ I created disc1, disc2, disc3 and disc4
/export and the subdirectories are chown nobody.nobody, chmod 777
Each of the fedora ISOs is loopback mounted to the appropriate directory

/etc/exports:
/export/fedora-core-2 *(ro,sync)

portmap was started before nfs

rpcinfo -p shows portmapper, status, nlockmgr, rquotad, nfs and mountd

ps aux shows portmap running

I believe I've gone through all of the posts on NFS on fedoraforum.org (found via a google search).

[crackers]

coyote4til7, I think your problem is the loopback mounts. If I remember correctly, you cannot export a filesystem if it's a "sub-mount" - in other words a filesystem that's mounted on another filesytem that's mounted - it has to be a primary mountpoint or an actual subdirectory of it. An NFS mount must have direct access/control over all subdirectories.

[coyote4til7]

That clears things up. I wasn't thinking about an iso being a different filesystem. Duh! What's really interesting is that Redhat 9 will do this -- I did it the week before last to set up a test box. Sometimes software that doesn't follow standards helps ;-} Oh well.

I did work around the problem I was having (creating a fedora install server). RedHat/Fedora has allowed you to use Apache as an install server. But, you had to do a franken-copy-n-paste from the ISOs into a new directory.

Sometime after RH9 (FC2 final is my guess), that's changed. Now you can just loopback mount the ISOs in a directory which Apache serves.

Apache's root on Fedora is /var/www/html -- below that I created installers/fedora. Then I created folders called disc1 to disc4 under that.

Next, you loopback mount each of the images:
mount -o loop [path to iso 1] /var/www/html/installers/fedora/disc1
Repeat with the other ISOs

Then, you'll need something to boot the machine you're installing FC2 on. In the disc1 folder, you'll now find the images folder. Burn boot.iso to a cdr or put diskboot.img on a jumpdrive.

Boot your machine off of the cdr/jumpdrive. Then you'll tell it your installing via HTTP and give it the IP of the server and the path to the discs (installers/fedora).

[sniperX]

Hi,

Since I use the graphic interface to setup NFS, it works on my FC2 Linux box. You can try it out form the following link.

http://fedoranews.org/dowen/nfsinstall/

[decals74]

i ended up fixing the problem by running this command on the fc2 nfs server:

mount -t nfsd nfsd /proc/fs/nfs

there is a new nfs-utils package out, not sure if it addresses this issue. i'm pretty surprised that fedora released fc2 with this bug. there are TONS of posts with ppl describing the exact same problem.

[crackers]

Huh! I haven't had any problems - of course, the NFS server is my wife's FC1 machine, so that may not be any indication...

[decals74]

hey crackers,

yeah i didnt encounter the same prob on my fc1 box, only fc2. i should read up more on nfs4 though before condemning fc2. mebbe its some new config crap that i've been overlooking.

that `mount -t nfsd nfsd /proc/fs/nfs` workaround came from an old fc1.91-test mailing list. it seems that nfs was a big prob with the beta.

its taken a bit of fiddling around to get it working as well as fc1 but i still love fc2. now if only root display on vnc would stop locking my box...