[1604] in SIPB-AFS-requests
Re: Cutover plans
daemon@ATHENA.MIT.EDU (Derek Atkins)
Tue Nov 15 19:08:25 1994
To: Greg Hudson <ghudson@MIT.EDU>
Cc: Derek Atkins <warlord@MIT.EDU>, sipb-afsreq@MIT.EDU, yandros@MIT.EDU
In-Reply-To: Your message of "Tue, 15 Nov 1994 18:55:31 EST."
<9411152355.AA00776@glacier.mit.edu>
Date: Tue, 15 Nov 1994 19:05:39 EST
From: Derek Atkins <warlord@MIT.EDU>
> It requires a change to the client CellServDBs, since rosebud is
> currently the first machine clients try to contact as a volume
> location server. If we do not change the client CellServDBs, then
> after the move, rosebud will not exist, and clients will time out.
No, this is not true. I am saying that we can leave the old rosebud
up and running along side the new rosebud, and things will be fine
until we decide to have clients update the CellServDB. In fact, the
new rosebud doesn't even have to run a DB server right away! So, as
far as clients are concerned, the files all just moved onto a
different machine (on a different network).
> If we do change the client CellServDBs, it requires coordinating with
> probe and dealing with outside sites. I think we should do this at a
> later time, and move both servers at once (leaving a volume location
> server behind on 18.70 for a short period of time).
This is fine -- we dont have to run a DB server on the new rosebud!
There is nothing that is waiting for the rs/6000 -- we could keep it
running an AFS db server.
I still think this is the best way to do this!
-derek
