[7118] in SIPB bug reports
Re: Dec 28 outages
daemon@ATHENA.MIT.EDU (Jonathon Weiss)
Fri Dec 25 16:48:41 1998
From: Jonathon Weiss <jweiss@MIT.EDU>
To: Brad Thompson <yak@MIT.EDU>
Cc: Jeremy Daniel <jdaniel@MIT.EDU>, sipb-machine-room@MIT.EDU,
bug-sipb@MIT.EDU
In-Reply-To: Your message of "Fri, 25 Dec 1998 13:41:31 EST."
<199812251841.NAA12358@drake.MIT.EDU>
Date: Fri, 25 Dec 1998 16:48:37 EST
> > athena% hesinfo sipb filsys
> > AFS /afs/sipb.mit.edu/project/sipb n /mit/sipb 1
> > AFS /afs/athena.mit.edu/contrib/sipb n /mit/sipb 2
> >
> > the sipb version is prefered...the read-write one is at
> > /afs/.sipb.mit.edu/project/sipb
> I was unclear. What I meant to say was that since the RW copy needs
> to be copied ober every night anyway, people should not notice that
> the RO copy takes time to update, whether the RO copy is in the sipb
> cell or the athena cell.
Actually, there are 4 copies of the sipb locker:
1) /afs/.sipb.mit.edu/project/sipb (SIPB cell RW)
2) /afs/sipb.mit.edu/project/sipb (SIPB cell RO)
3) /afs/.athena.mit.edu/contrib/sipb (Athena cell RW)
4) /afs/athena.mit.edu/contrib/sipb (Athena cell RO)
Every night ronald-ann copies (with synctree) from 1 to 3 and then
releases (with vos) from 1 to 2.
Also every night persephone releases from 3 to 4 (with vos) along with
releasing a bunch of other volumes in the Athena cell that we don't
care about.
The issue is that if persphone runs first, there might be an extra
days worth of delay getting stuff from the canonical location (1) to
the prefered location for users (since it would be 4 instead of 2).
That said, I don't know what all of the relevant times are offhand, so
I don't knwo if this is an issue, but over winter break, I don't think
it matters much. Besides, Karl or I can always release volumes if
there is an emergency.
Jonathon
