Bernd Schubert wrote:
> On Tuesday 17 April 2007 09:27:36 Patrick Begou wrote:
>> Fajar Priyanto wrote:
>>> Hi all,
>>> I'm setting up a 2-node cluster for httpd, pure-ftpd, and nfs failover.
>>> All seems ok. httpd, pure-ftpd, nfs can failover in all tests (unplugged
>>> eth0, reboot, shutdown, etc).
>>>
>>> The problem is, the nfs is exporting a partition on each server without
>>> shared storage, in this case it's exporting /var/ftp/pub softlinked to
>>> /home/pub
>>>
>>> The problem arises when failover occurs, the client mounting that export,
>>> has to remount it, because of this error:
>>> cannot access /mnt: Stale NFS file handle
>> Is it a read-only access for the clients ? If not, how do you
>> synchronize them ?
>> Did you try to use upd nfs instead of default tcp (proto=udp option on
>> the clients)?
>
> This won't help.
>
>> But I think you will have to share /var/lib/nfs between the two HA
>> servers...
>
> If you are using a 2.6.x kernel with nfsd mounted to /proc/fs/nfsd this is in
> principle not neccessary anymore. Nowadays its only required for statd data,
> so file locking.
>
> Alan is right, one needs an identical image. (Well, if you KNOW what you are
> doing, it doesn't need to be entirely identical. At least the fsid, root
> inode number (of your exported dir) and inode generation number MUST be the
> same.)
No, more than that.
EVERY SINGLE INODE has to have the same number.
Sometimes, you get away with less, but unless this is true, someday
you'll get burned.
--
Alan Robertson <[EMAIL PROTECTED]>
"Openness is the foundation and preservative of friendship... Let me
claim from you at all times your undisguised opinions." - William
Wilberforce
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
