[OpenAFS] Volume backup in distinct server

[Derek Atkins]

Joao Pedro Barreto <jpbarreto@gsd.inesc-id.pt> writes:

> Please correct me, but if I use 'vos addsite', I can only create a
> readonly volume of my read/write volume (when I tried 'vos addsite'
> with the readonly clone, it wasn't allowed). This means that, issuing
> a 'vos release', my read/write volume will be locked for some time,
> which is not desirable.
> Am I correct?

I do not think you are correct.  I believe the process is:

lock vldb entry
lock RW
clone to local RO
unlock RW
copy to other sites
unlock vldb entry

So you should have complete RW access after the local cloning.

-derek

> Using the following approach, I guess I could avoid this problem:
> create a backup volume on the same server/partition of the original
> W/R volume, then making 'vos dump Myvolume.backup | vos restore
> backupVol' to transfer the contents to a volume on the other
> server/partition.
> Do you agree?
> 
> Joao
> 
> Derek Atkins wrote:
> > "vos addsite"
> > "vos release"
> > -derek
> > Joao Pedro Barreto <jpbarreto@gsd.inesc-id.pt> writes:
> >
> >>Hi,
> >>
> >>But, after creating the clone in the same server/partition as the
> >>original volume, how can I copy its contents accross the net to the
> >>other server/partition?
> >>Is there any way I could do this using exclusively AFS commands? Or do
> >>I have to copy everything by my own hands, for example using a simple
> >>Unix cp command? And how can I configure it to be periodic, for
> >>instance, to be performed daily?
> >>
> >>The best thing I can think of is to issue a 'vos dump' command to the
> >>readonly clones, and piping its output to a 'vos restore' on the other
> >>server/partition. Putting this line for each volume I want to backup
> >>in a shell script and configure it in chron, I might have my problem
> >>solved.
> >>Do you find any problems with this solution? Or is there any better
> >>approach?
> >>
> >>Thank you.
> >>Joao
> >>
> >>
> >>Derek Atkins wrote:
> >>
> >>>Hi,
> >>>The backup volume is meant for user access to backup data, not for
> >>>"replication across servers".  It's on the same server to reduce
> >>>the amount of disk space that is used.  It's purpose is not to protect
> >>>against lost disks, it's purpose is to protect against "rm".
> >>>What you seem to want is ReadWrite replication, which is something
> >>>that AFS does not do.
> >>>One thing you CAN do is make a RO clone on the same server/partition
> >>>as the RW (as well as on the other server).  The local RO clone wont
> >>>take up much disk space (it's a clone after all, just like the Backup
> >>>volume), but it does help in terms of the locking.  The RW only gets
> >>>locked in order to make the clone, and then the clone is copied across
> >>>the net.  This means your RW volume isn't locked for the whole release
> >>>process, only during the clone.
> >>>-derek
> >>>Joao Pedro Barreto <jpbarreto@gsd.inesc-id.id> writes:
> >>>
> >>>
> >>>>Hi,
> >>>>
> >>>>I administer an AFS domain which has two servers storing file volumes.
> >>>>I wanted to configure volume backup, so that each volume stored in one
> >>>>server's disk would be replicated (or backed up) in the other server's
> >>>>disk.
> >>>>My first approach was to create read-only replicas of each volume. For
> >>>>instance, if a volume was stored in server A, I would create a
> >>>>read-only volume for it on server B. However, I soon discovered that
> >>>>the release of volumes (which I'd perform daily in order to update the
> >>>>read-only volumes) NEEDS TO LOCK the original volume, which can be a
> >>>>serious drawback if we consider that most volumes are sufficiently big
> >>>>to take some hours to be released.
> >>>>I then found about the backup functionality which, according to AFS
> >>>>documentation, should be the best choice for my problem. Mainly, this
> >>>>solution doesn't require locking of the original volumes to update the
> >>>>backup volumes. However, this option only allows creation of backup
> >>>>volumes in the SAME PARTITION (AND SERVER) as the original
> >>>>volume. Which, of course, is insuficient to my requirements of having
> >>>>backup volumes in distinct computers from their original volumes (so
> >>>>that, if a computer fails and its data are lost, the volume can still
> >>>>be obtainable from the other computer).
> >>>>
> >>>>Does anyone can help me on this? how should I solve my problem?
> >>>>
> >>>>Thank you in advance.
> >>>>
> >>>>Joao
> >>>>jpbarreto _AT_ gsd _DOT_ inesc-id _DOT_ id
> >>>>
> >>>>
> >>>>_______________________________________________
> >>>>OpenAFS-info mailing list
> >>>>OpenAFS-info@openafs.org
> >>>>https://lists.openafs.org/mailman/listinfo/openafs-info
> >>>
> >>
> >>
> >
> 
> 
> 

-- 
       Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
       Member, MIT Student Information Processing Board  (SIPB)
       URL: http://web.mit.edu/warlord/    PP-ASEL-IA     N1NWH
       warlord@MIT.EDU                        PGP key available