[1684] in SIPB-AFS-requests
Tonight's outage
daemon@ATHENA.MIT.EDU (ghudson@MIT.EDU)
Tue Dec 27 04:07:05 1994
From: ghudson@MIT.EDU
Date: Tue, 27 Dec 1994 04:05:38 -0500
To: star-maintainers@MIT.EDU
Cc: sipb-staff@MIT.EDU, sipb-prospectives@MIT.EDU
We took down all of the machines in the machine room and reorganized
them starting at midnight tonight. The desks now look like:
---- a - bloom-picayune with a head
|k | b - bloom-picayune's two 1GB disks
---- c - ronald-ann2 and rosebud2 with a head
---- d - rosebud2's 4GB and 2GB disks
| | e - ronald-ann2's four 1GB disks
|j | f - opus with a head
---------- ---- g - opus's 1GB disk
----| fg hi |---- h - bloom-beacon with a head
|de|----------|m | i - bloom-beacon's 1GB disk
|c | |l | j - bed with two 300MB and two 1GB disks,
|b | |n | k - bed's head and keyboard on a stand
|a | | | l - penguin-lust with a head
---- ---- m - penguin-lust's 1GB disk
------ ---- n - anxiety-closet
| q | |m | o - charon with a head, on a cart
------ ---- p - charon's 300MB disk
---- q - a VT100 on a cart
|o |
|p |
----
Apart from looking like something from Tron, this arrangement lets us
get at all of our machines but bed from the middle of the U shape. It
also means that all of the cabling is exposed and behind the desks,
unlike the old arrangement where we had two sets of machines back to
back. We are using three underground power circuits and five power
strips; two of the power strips are dedicated to outgoing machines
(one to bed, one to charon and rosebud). The 18.181 Delni is
underneath beacon, and the 18.70 Delni is underneath anxiety-closet.
We economized on ethernet cables a bit by not needing as many
extension cables; we needed one for anxiety-closet and one for opus
(temporary until it moves back to 18.181).
Note that opus has a 1GB disk now, and the AFS servers didn't gain a
disk. I think we mostly decided that we don't need an extra gigabyte
of AFS disk right now, and would rather do the news cutover sooner.
In fact, after we're done with cutovers, we will probably leave the
extra 1GB of disk floating around until we actually need it for
something (such as the new rtfm, or if we run short on AFS space
sooner than we expected).
We had trouble with both ronald-ann2 and rosebud2 at boot time.
rosebud2 failed to recognize it's 4GB disk as it has been prone to do;
I believe the solution to this is to turn on the 4GB disk, wait a
minute or so, and then turn on rosebud2. Unfortunately, power outages
won't be quite so disciplined, but I can't think of any good
solutions. ronald-ann2 displayed a cursor and didn't do anything
until I powered everything off and on again; I don't think we've seen
anything like that before.
charon also failed to boot because it was dropped on the floor while
we were moving things around. It hung on an internal SCSI error.
Swapping part of the case with hodge solved that problem, and charon
is now up as of 3:55am (so it was down for four hours, an hour longer
than scheduled).
Other than charon, all services were down for about two and a half
hours, which was on schedule. We got a couple of questions on -i
help, and one message sent to postmaster@mit.edu from a network
provider asking if bloom-picayune was down, but otherwise it was
pretty quiet.
