Sender: |
|
Subject: |
|
From: |
|
Date: |
Mon, 13 May 1996 18:18:52 +0200 |
In-Reply-To: |
Message of Fri, 10 May 1996 14:51:37 EDT from LISTSERV
give-and-take forum < [log in to unmask]> |
Reply-To: |
|
Parts/Attachments: |
|
|
On Fri, 10 May 1996 14:51:37 EDT Stan Horwitz <[log in to unmask]> said:
>I was just told by one of the people who runs the system on which we run
>Listserv that if we were to shut Listserv down to back it up, it would
>be down for several hours
Uh? What is your backup device, a 3410-1 with a tape mount in between and
the operator watching a movie that's so good he can't get away from the
TV until it's over? :-) LISTSERV 191 is typically 20-30 cyl at large
sites, it should only take a couple minutes to back up. Alternatively you
could create a shadow 191 minidisk and have LISTSERV copy the files over
once a day, then release and detach the disk. You could then back up this
alternate minidisk instead of the 191.
In general the problem of making reliable backups on a live system is
unsolved. The system that comes closest is OSF/1, the backup program can
tell the system to take a snapshot of an entire partition so that the
backup program sees the files as of this instant regardless of any
further change (a bit like SFS, but for the entire file system). But not
even that ensures that the snapshot is in a stable state. An application
could be in the process of appending to a file, etc. Any application that
can recover automatically from a power outage should recover from
restoration of such a backup, but ordinary end users seldom write
applications this way :-) As far as I know, OSF/1 is the only system that
can do that. I hope and assume that TNFS will also have this ability
under VMS.
Eric
|
|
|