[258] in Athena_Backup_System
ABS minutes for 6/13/96
daemon@ATHENA.MIT.EDU (Jonathon Weiss)
Thu Jun 13 16:43:26 1996
Date: Thu, 13 Jun 1996 16:45:36 -0400
To: athena-backup@MIT.EDU
From: jweiss@MIT.EDU (Jonathon Weiss)
attending: jweiss, wdc, vrt, delgado, bwmelans, miki, dkk
There is a web server on metamucil. jweiss is concerned that this
happened without any discussion especially WRT security. Diane and
Bill explained that this was just a test to see how it works.
Jonathon is satisfied aslong as there is a discussion at an ABS
meeting, before this is run on a production machine. Diane wants a
document listing how to set up a secure web server. Dave and bill
comment that this may be impractical for several reasons. (web server
security is often applicable to a specific web server not to all
servers. We aren't sure that anyone around here is likely to be able
to write this and keep it up to date.)
Req's:
Dave has a question about the wording of req. about the load on the
AFS servers. This should include DB servers as well as the file
servers in the case of AFS. other non-ABS servers too. Bill will
wordsmith.
Diane: req about tape to partition maping doesn't exist why/when/how
did this end up in the req's? Bill will change this, but asks what if
we want to restore a partition. Diane: the db can derive this, I
don't want the db to be required to have it. Given that volumes can
move we may have to generate the volume to partition mapping from AFS.
We can tell ABS to sync from the vldb and then do the partition
restore.
in crippled mode backing up a server or partition is required to get
the list of volumes from the servers, so we don't miss any new
volumes. (bill will remove the requirement that confused many team
members (it was #7) modify req #2 and add a new #3, in order to have
this all make sense.)
dumping the clone in cripple mode will *not* cause the volume to be
re-cloned.
Bill gives an update about the req's. and talking to Jeff, etc.
Jonathon
