[1682] in SIPB-AFS-requests

home help back first fref pref prev next nref lref last post

ronald-ann converted to ronald-ann2

daemon@ATHENA.MIT.EDU (ghudson@MIT.EDU)
Mon Dec 26 17:08:07 1994

From: ghudson@MIT.EDU
Date: Mon, 26 Dec 94 17:07:55 -0500
To: sipb-afsreq@MIT.EDU


Derek, Matt Braun, and I backed up ronald-ann's /var/usr/local and
/etc/crontab onto rosebud2's /var/usr/local, brought down ronald-ann,
and reinstalled it as ronald-ann2 on 18.181.  Matt created a srvtab,
and we also have (courtesy of probe) an rcmd srvtab for ronald-ann to
make /var/local/sync/synclocker work.  We restored /var/local and
/etc/crontab from before the reinstall, and brought
/etc/athena/rc.conf into conformance with rosebud2's.  I fixed
/etc/crontab and /var/local/sync/release.sh to refer to
/var/local/sync instead of /var/usr/local/sync, and modified
/var/local/sync/release.sh to release volumes on ronald-ann2,
rosebud2, and rosebud.

Matt Braun updated the Athena CellServDB to read:

	>sipb.mit.edu             #MIT/SIPB cell
	18.181.0.23                     #ronald-ann2.mit.edu
	18.181.0.22                     #rosebud2.mit.edu
	18.70.0.210                     #rosebud.mit.edu

This will prevent clients from timing out on ronald-ann.  After we
move the data from rosebud to ronald-ann2, we can change the namespace
and Athena CellServDB at the same time to move the names "ronald-ann"
and "rosebud" to 18.181 and assign a name like "old-rosebud" to
18.70.0.210.  At this point, the Athena CellServDB will read:

	>sipb.mit.edu             #MIT/SIPB cell
	18.181.0.23                     #ronald-ann.mit.edu
	18.181.0.22                     #rosebud.mit.edu
	18.70.0.210                     #old-rosebud.mit.edu

This should not affect service (as long as ronald-ann2 and rosebud2
remain cnames for the 18.181 machines for a period of time).  I think
that if we use "hostname" to change the machines' hostnames and
restart a few processes like syslog, we should be able to avoid
rebooting the AFS servers at this time.

We are doing an AFS backup for most of the period of time between now
and midnight, and won't be moving volumes during that time.  We should
be able to do that tomorrow, though.


home help back first fref pref prev next nref lref last post