Bernd Schubert wrote:
> On Wednesday 18 April 2007 16:38:46 Alan Robertson wrote:
>>> 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.
>
> Well, this gets a bit out of topic and should be better transferred to the
> nfs
> mailing list ;)
>
> The nfs server will return ESTALE if the inode/inode-generation number
> doesn't
> match. Now the nfs client is supposed to through away its nfs-fh and to
> request a new fh from the server - it has to assume the old file/directory
> has been deleted. Applications having opened this file will get an error, of
> course. In most cases this is a problem for you, but there are also
> situations when this is acceptable. (Thatsway I wrote you have to know what
> you are doing).
>
> Requesting new nfs-filehandles can be done recursively down to the root-fh,
> which is IMHO the only filehandle that has to stay as it is.
"Applications having opened this file will get an error"
So, this is exactly what I was talking about. In the overwhelming
majority of cases (99+%), the application isn't NFS-aware, and this will
stop it dead.
Now, if you're the application developer and you ONLY run one
application that you wrote yourself, you could take care of this.
But, the _administrator_ definitely cannot take care of this problem
themselves, except by the method I described - making every single inode
match.
--
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
