acad-cl5 crashed for the second time in two weeks at 18:45 june 8th. It was restored to service at 08:20 following some testing.
acad-cl5 crashed at approximately 1635 and was restored to service about 10 minutes later.
acad-cl3 crashed friday at 1700. Was restarted at 0930 saturday.
acad-cl5 and acad-cl6 have been reved to fedora core 5 with the appearance of success which means we're going to want to take a 3-4 hour outage on each of the remaining machines in the next couple days.
acad-cl5 and acad-cl6 slated to be upgraded tomorrow to Fedora Core 5. if everything goes well the remaining nodes will be upgraded in the following days.
acad-cl2 has been returned to service.
/scratch is now back.
The swap was sucessful... There are still a few cleanup issues, notably /scratch will not be remounted until the rsync finishes, hopefully before 1700. At this time all the licensed software managers should be working.
acad-cl0 will be swapped out at 12:29. this will require a reboot of all nodes, wall announcement was sent out starting at 0900. post reboot the scratch filesystem will not be available immediatly until it finishes syncing.
Something is wrong with acad-cl2 perhaps a bad disk, triage is tentatively scheduled for monday Febuary 20. The problem with the acad-cl0 may or may not have been isolated, It will take a couple hours to tell. The patience of all the users involved is appreciated.
acad-cl0 crashed at 17:15 PST. Problem may be heat related, outage won't be over until a reboot can be performed probably around 0745 Monday the 13th. Further attempts at troubleshooting will probably require a restart and move acad-cl4, later in the morning. If you notice issues after 0900 on the 13th please email joelja@uoregon.edu and consult@uoregon.edu.
The new replacement for Darkwing is becoming available to early adopters. The host is currently known as geoduck but will soon adopter the moniker shell.uoregon.edu. Users with a uoregon.edu account should be able to log in at this time. The machine offers 4 dual core opteron 870 processors and 8GB of ram. It is subject to the same service schedule as darkwing and should not be considered in full service yet.
acad-cl1 and acad-cl2 have been upgraded with donor cpu's from acad-cl5 and acad-cl6, They also have been upgraded to 4GB of ram. Another maintenance outage is not immediatly forseen.
acad-cl5 and acad-cl6 have been upgraded with Opteron 250 (2.4ghz) cpu's replacing the 1.8ghz cpu's previously present. They also have been kicked up to 4GB of ram each. next weekend upgrades for acad-cl1 and acad-cl2 are planned.
Replacement (faster) precssors for acad-cl5 and 6 have arrived. I am tenatively scheduling an outage for Saturday the 14th to effect the swap. The net result should be 4 4 machines with faster processors and 4 more with 4GB of ram.
Ethernet interfaces on private network now configured for 9000 mtu to improve NFS performance.
Rack maintenance inadvertantly caused a power outage across most of the nodes. sorry for any inconvenience caused as a result.
acad-cl3 became non-responsive due to a run-away program at 10:59:34. It was rebooted. Chewing through 4GB of swap is bad. Acad-cl2 was rebooted monday because no-one was using it. As machines get rebooted they are coming up on a new kernel.
MaCaulay2-0.9.2 and radiance-3R7P2 installed on all nodes Mathematica upgraded from 5.0 to 5.2 on all nodes
MaCaulay2-0.9.2 and radiance-3R7P2 installed on acad-cl1.
IRAF v2.12 and associated STSDAS and TABLES modules have been installed.
acad-cl4 back on 1.4ghz cpu's and seemingly functional.
The new cpus in acad-cl4 appear to just not be stable, we will swap them back out this weekend.
acad-cl4 was found to have crashed this morning and was restarted. acad-cl5 was rebooted after 80 days to run a newer kernel.
Upgraded matlab from Version 7.0.1.24337 (R14) Service Pack 1 August 20, 2004 to Version 7.1.0.183 (R14) Service Pack 3 August 02, 2005. Binary is /usr/local/bin/matlab.
SAS 9.1 was also installed. It can be invoked as sas_en. the directory /usr/local/sas/sas_9.1/bin was added to the system-wide bash-profile, if you use another shell you'll have to add it.
acad-cl0 the nfs server will be rebooted Friday October 7th This will cause a brief interuption in disk service on the other nodes. No data should be lost.
acad-cl4 now has 1.8ghz opterons instead of 1.4ghz opterons.
acad-cl4 will be restarted tomorrow morning to perform some system maintenan\ce.
Reboot needed on acad-cl1, scheduled for 0930 PDT Tuesday.
acad-cl1 was found to be non-responsive to login attempts rebooted at 0818. Upgrade to Fedora core 4 is complete across all nodes.
Acad-cl2 to be upgraded to fc4 on tuesday @ 0930. home11-16 mounted, now undergrads have access to the cluster hosts as well.
Acad-cl3 upgraded to fc4 leaving cl-1 and cl-2 still to be done.
Acad-cl6 is back, some software is still to be installed. The R-system has been upgraded to version 2.1 on acad-cl4-6.
Acad-cl6 is in the process of rebuilding, it should be back by tomorrow.
Acad-cl6 is out with a disk failure.
acad-cl4 was temporarily unavailable this morning.
acad-cl4 is back, now running FC4. acad-cl4 is the first of the machines to be upgraded to 4GB of ram. Expect acad-cl1-3 to be upgraded to FC4 within the next few weeks.
acad-cl4 will be down till monday morning due to hardware upgrade failure.
Equipment failure in 180 CC caused an outage from approximately 18:07 to 19:30 when service was restored. acad-cl1 presently is wedged with a failed disk. nodes cl-2 through cl6 are operational.
Home directories have been remounted off the netapp this morning.
also
The hosts acad-cl5 and acad-cl6 are now available. They are dual opteron 244 (1.8Ghz) machines with 2GB of dual-channel ddr400 ram. Compared to acad-cl1-4 They have maybe 30% more cpu and 250% higher peak memory bandwidth.
/usr/bin/R. For more information on the R system visit:
http://www.r-project.org/
online access to the manuals is here.
/usr/local/bin/matlab