Skip to contentSkip to conversation

can't modify /etc/exports without blowing away existing clients

DISCUSSION John Z. Bohach 4 participants 4 messages Feb 16, 2009 – Feb 17, 2009 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hello,

When I modify /etc/exports and re-run exportfs -a -r, newly added/removed/modified entries take effect, EXCEPT that all existing clients lose their current mounts. Existing clients have to umount then mount NFS filesystems that were not even modified.

Is this intentional behavior? If I try to manually add a new directory for export, and just use "exportfs -i ..." and specify the new directory to export that way, all existing entries in /etc/exports are gratuitously UNexported, all clients lose their current mounts, and only the one entry specified via the command line of "exportfs -i ..." is exported. This doesn't seem like intentional behavior? Or is it just the exportfs program that's broken?

I'm running linux-2.6.24.4 and using exportfs from nfs-utils-1.1.3 though I've noticed this behavior on older versions of both.

Is there some NFS option I'm supposed to use in /etc/fstab on the clients to keep this remount thing from being necessary? I'm pretty sure I tried both soft and hard mounts. I'm currently using TCP as the mount transport layer...would UDP fix this issue, I haven't tested that...either way, even with TCP, it would be a big blow against using TCP as the mount transport if umount/mount sequence was required on the clients everytime time the server re-ran exportfs, which is what is happening now.

Anybody know what's going on?

Thanks,

John

P.S.: Please keep me in the CC list, I'm not subscribed to the list.

A message that is not in the archive (sent off-list or not yet received): [email protected]

#2 lore raw
Quoted 27 lines · John Z. Bohach

Hello,

When I modify /etc/exports and re-run exportfs -a -r, newly added/removed/modified entries take effect, EXCEPT that all existing clients lose their current mounts. Existing clients have to umount then mount NFS filesystems that were not even modified.

Is this intentional behavior? If I try to manually add a new directory for export, and just use "exportfs -i ..." and specify the new directory to export that way, all existing entries in /etc/exports are gratuitously UNexported, all clients lose their current mounts, and only the one entry specified via the command line of "exportfs -i ..." is exported. This doesn't seem like intentional behavior? Or is it just the exportfs program that's broken?

I'm running linux-2.6.24.4 and using exportfs from nfs-utils-1.1.3 though I've noticed this behavior on older versions of both.

Is there some NFS option I'm supposed to use in /etc/fstab on the clients to keep this remount thing from being necessary? I'm pretty sure I tried both soft and hard mounts. I'm currently using TCP as the mount transport layer...would UDP fix this issue, I haven't tested that...either way, even with TCP, it would be a big blow against using TCP as the mount transport if umount/mount sequence was required on the clients everytime time the server re-ran exportfs, which is what is happening now.

Quoted 1 line

Anybody know what's going on?

Does the command 'grep nfsd /proc/mounts' show the 'nfsd' virtual filesystem as being mounted on /proc/fs/nfsd?

Cheers
Trond

#3 lore raw
In reply to a message not in the archive
Quoted 8 lines · John Z. Bohach

Hello,

When I modify /etc/exports and re-run exportfs -a -r, newly added/removed/modified entries take effect, EXCEPT that all existing clients lose their current mounts. Existing clients have to umount then mount NFS filesystems that were not even modified.

Is this intentional behavior?

No, that shouldn't happen. As trond says, check whether you have the "nfsd" filesystem mounted.

Quoted 7 lines

If I try to manually add a new directory
for export, and just use "exportfs -i ..." and specify the new directory to export that way, all existing entries in /etc/exports are gratuitously UNexported, all clients lose their current mounts, and only the one entry specified via the command line of "exportfs -i ..." is exported. This doesn't seem like intentional behavior? Or is it just the exportfs program that's broken?

I didn't know -i; looking at the man page, I think that's may be what's intended. (I haven't looked at the code.) Why not just drop the "-i"?

--b.

Quoted 23 lines

I'm running linux-2.6.24.4 and using exportfs from nfs-utils-1.1.3 though I've noticed this behavior on older versions of both.

Is there some NFS option I'm supposed to use in /etc/fstab on the clients to keep this remount thing from being necessary? I'm pretty sure I tried both soft and hard mounts. I'm currently using TCP as the mount transport layer...would UDP fix this issue, I haven't tested that...either way, even with TCP, it would be a big blow against using TCP as the mount transport if umount/mount sequence was required on the clients everytime time the server re-ran exportfs, which is what is happening now.

Anybody know what's going on?

Thanks,

John

P.S.: Please keep me in the CC list, I'm not subscribed to the list. --
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#4 lore raw
Quoted 2 lines

No, that shouldn't happen. As trond says, check whether you have the "nfsd" filesystem mounted.

Thanks...got it resolved...squirrly startup script was mounting nfsd fs just after it started rpc.statd...

--john