[OpenAFS] Restore primary from replica server

Kim Kimball dhk@ccre.com
Wed, 11 Jul 2007 12:35:44 -0600

AFS doesn't require a 'primary server' nor does it require segregation 
of RW volumes and RO volumes.

It will help to know what makes a server your 'primary.'

That being said, some sites do consider a specific AFS file server to be 
their 'primary.'  This may be for non-AFS reasons.  For example, some 
sites have kept RW data on beefier (faster/higher availability/whatever) 
hardware, and want all RWs kept on that hardware.  If for this or some 
other reason you've segregated RW volumes from RO, by all means restore 
to your 'primary.'

In general a volume can be restored to any server/partition with 
sufficient space.

There are different types of restores.  Two that come to mind are 
'restore to replace corrupted volume' and 'restore to retrieve file I 
just mangled beyond hope.'

To replace a corrupted volume I can restore over an existing volume 
(overwrite in place using the same volume name and location) if I so 
choose -- this is generally easier because I don't have to remove the 
corrupted volume and don't have to  finagle the VLDB entry, and won't 
get complaints that "a volume of this name already exists," etc.

I can restore without overwriting existing volume, for example to 
retrieve some subset of files/directories from time X.  (different name, 
location as convenient.)

As for being 'standard' to restore data to the primary?  I don't, others 

I never taught this as best or even recommended practice when teaching 
for Transarc/IBM or later when teaching my own classes, but your 
hardware configuration or some other factor may make it relevant to your 
AFS installation.

Kim Kimball

Jeff Greer wrote:
> Is it standard to restore data to the primary AFS server from a read 
> only replica server?
> Thanks.
> Jeff
> _______________________________________________
> OpenAFS-info mailing list
> OpenAFS-info@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-info