SP/2 Changelog
A record of what I've done to our SP/2.
I've also recorded things which have occured,
so I can keep track of oddities.
I started this after I started to change things.
Oops.
I occasionally remember things that I've done and
record it in the A long time ago section.
Mon May 20 16:07:58 CDT 1996
- Install FixDist an IBM utility for installing PTFs.
Fri May 17 11:58:39 CDT 1996
- Create a raw logical volume, disk2, on hdisk2. I created
the device writable by group paradise device as
/io/rdisk2.
Thu May 16 11:45:59 CDT 1996
- Cleanup /tmp on the nodes.
- Increase size of /var to 8MB; the previous 4MB was
too close for comfort with 99% to 100% utilization.
- Reseat the SCSI and Switch interface cards in sp2-03.
The boards were loose. This fixed the switch problem,
but not the SCSI problems.
It may be time to call IBM about this loose board problem.
Wed May 15 15:16:06 CDT 1996
- Power-cycle sp2-03 so its outboard disks will configure.
That didn't work.
Disk Shuffle
hese fall under the category of The Disk Shuffle,
which I've been meaning to do for some time now.
- Remove the /home.old filesystem.
- Remove the jetty filesystem and logical volume
- Remove the beach logical volume
- Remove the miscvg volume group (hdisk1 was miscvg)
- Attach hdisk1 to rootvg.
- Add /local, a 1G filesystem.
Thu Mar 21 13:42:33 CST 1996
- sp2-16 has problems with its microchannel bus. Either
the bus or the cards in it are loose.
Both the SCSI and the Switch cards are loose, and they wouldn't
configure.
I wiggled the cards from the backside to reseat them.
This may explain the earlier problems with this node.
- Todo: Call IBM to fix the microchannel bus.
- Fetch and install monitor, an AIX top and systat
equivalent.
It's installed in /usr/local/bin/monitor.
- I configured the quantum at SCSI address 4
on the external bus, hdisk4, as
a JFS filesystem. The volume group for the drive is vg4.
The filesystem is mounted as /shore/disk4
Fri Feb 9 13:09:50 CST 1996
- I power-cycled sp2-16. It was not reachable
from the (ether)net.
- I rebooted a bunch of nodes. Cron had run wild
on them. Interestingly, it didn't run while on sp2-01,
where amd had screwed up.
Mon Jan 29 14:03:40 CST 1996
- Tim installs afs-3.4a on the CW. This must be
done before the AFS servers can be upgraded.
Sat Dec 30 20:32:58 CST 1996
- Debug switch problems. The switch was turned off, and needed to
be restarted. I think that rebooting the -01 node disabled it.
- Debug automounter problems on nodes -01 and -04. The
sp2-cw:/home filesystem is mounted, but the link from /u to the
mountpoint was missing.
Fri Dec 29 20:55:16 CST 1995
- I changed /etc/environment on the nodes to make
Central Standard Time the system timezone.
Thu Dec 28 14:44:08 CST 1995
- Change karthik's shell to /usr/local/bin/tcsh.
- Add /usr/local/bin/tcsh to /etc/security/login.cfg on CW and nodes.
- Distribute karthik's version of tcsh in /usr/local/bin.
- Create /usr/local hierarchy on all nodes.
Wed Dec 20 12:20:15 CST 1995
- Karthik discovered bunches of crons on some
of the nodes. I suspect they may be hung in nfs-mounts,
but I'm not sure (no evidence). I rebooted all nodes as an
expedient measure.
Fri Dec 15 16:30:16 CST 1995
- The jfs filesystem /dev/jetty is mounted as /shore/logs.
- The raw filesystem /dev/beach has a symbolic link
from /shore/volumes/beach.
- I created a /shore directory structure on all
the nodes. This contains logs and volumes subdirectories,
which will contain links or mounted filesystems as approriate.
- Created a jfs filesystem, on miscvg, /dev/jetty on all
the nodes. This filesystem is used for shore logs.
- I created a raw filesystem on miscvg, /dev/beach, on all the
nodes. This partition has a 1 gigabyte, and is for the use of
the paradise group. The raw device is group+rw paradise.
- Created a volume group miscvg on all the nodes, which
contains hdisk1.
A long time ago, in a galaxy far, far away
- I changed /etc/environment on the CW to make
Central Standard Time the system timezone.
- On all the nodes, I renamed the /home filesystem to /home.old.
/home is now a link to /u. This allows the filesystem namespace
on the nodes to be the same as that on the CW (/u and /home identical).
I needed to do this to allow poe to work.
- I created a new /home filesystem on the CW, using space
from hdisk1. The space from Hdisk1 is in uwvg.
- With help from tim
I installed the AFS client software on the CW. All users are
now authenticated through AFS.
SP/2 Information
Bolo's Home Page
Last Modified:
Fri Dec 29 20:53:30 CST 1995
bolo (Josef Burger)
<bolo@cs.wisc.edu>