[69] in Athena_Backup_System
Re: error handling
daemon@ATHENA.MIT.EDU (dkk@MIT.EDU)
Thu Jan 26 12:58:57 1995
From: dkk@MIT.EDU
Date: Thu, 26 Jan 95 12:58:28 -0500
To: jweiss@MIT.EDU
Cc: athena-backup@MIT.EDU
In-Reply-To: <199501140118.UAA15345@the-other-woman.MIT.EDU> (message from Jonathon Weiss on Fri, 13 Jan 1995 20:18:52 EST)
(Sorry, I didn't think of this the first time I read your mail.)
> after machine crash
> This can be looked at as a special case of the process crash below
>
> after process crash
> Inform master slave is back, return logged errors, but not successes,
> abort dump.
When the machine boots, devices will be reset and configured. The
state of some storage devices, such as tape and WORM (write once, read
many) drives, may depend on device specifics. So the tape slave
should either do the safest thing (abort or start over) or it needs to
be able to tell the difference between the start-over cases and the
keep-going cases.
I'm not sure how best to deal, but I'd be inclined to always abort,
because I think there's going to be a lot of complexity in figuring
out when it's safe to keep going with the dump set. Imagine a crash
caused by a power glitch which also takes out the tape jukebox -- does
the tape get re-filed upon power-up?...
