[OpenAFS] Partial replication

[Charles Clancy]

> Unfortunately I don't have enough disk to replicate my WHOLE AFS space
> to these machines. I was thinking that I could just replicate the
> root.* volumes and the 'user database'...

So, these additional machines will be db and file servers?  They don't
necessarily have to be both.  Since you're not using kaserver, you'd
probably want to run the vlserver and ptserver on them.  Then running the
file server too, you could replicate any volumes you feel are critical.

Unlike kerberos and ldap, these machines will be used all the time, not
just in the event that the primary fails.

> The idea was that I could get the main AFS system up and running from
> these machines if the main AFS system crashes (for some reason).

Since these machines would only be lacking the extra volumes, you could
restore them from tape, and be up and running.

> Since I'm doing backups every day, it should then be possible to put
> the backups back on disk...

If you did crash, and expected the main server to be down for a while,
you'd probably want to:
1. remove the downed server from the cellservdb (possibly)
2. sync the vldb on the secondary server(s) so they only know about their
own volumes (not the ones on the dead server).
3. restore the missing volumes from tape to the working server(s).

> Is there anything else that I could do to ease the 'restoration' in case
> of a crash?

In my opinion, it's best to have enough volumes replicated to the other
servers as is necessary to keep things going while the primary server is
repaired.  Mucking with the vldb and restoring from tape is not fun and
should really only be done if you expect your primary server to be down
for a long while.  Think of them as providing minimal services for a
temporary period, rather than completely taking over.

[  t charles clancy  ]-[  tclancy@uiuc.edu  ]-[  uiuc.edu/~tclancy  ]