- Put a copy of the EQ Experts database on the Pasadena Internal web page at
http://insider.gps.caltech.edu/eqexperts/
- Added Anthony to paging.
- Talked with Carl, Mike, and Mandy about reorganizing and cleaning up the
telemetry room and its wiring.
- Had ITS remove the MAC address block on JD's computer on Tuesday.
- Carl installed a new shelf in the Spring rack, and we moved the temp
wavepool disks on to it.
- Attended the web planning meeting in Menlo Park on Wednesday and Thursday.
- Made a patch for the Simpson maps so that the hour in the local time on
the event pages is always displayed with two digits. There was an event
at 00:02:55 on Friday morning, and the event page had the time as '0:02:55'.
Sent the patch to Bob Simpson.
- Fought with Pluton some more. It can boot successfully, but whenever Hotspot
starts doing NFS to it, it locks up.
- Moved Granite and Rtdev to different shelves on Tuesday.
- Wrote to SDSC to ask about their experiences with the 3ware cards. They said
that they had had the same lockup problems, and had switched to Promise
UltraTrak RAID enclosures instead. Called Promise and got some information
about these. They have not been tested with Solaris, but they have an
evaluation program.
- Rift crashed mysteriously at about noon on Wednesday.
- Moved Ehzsouth, Agent86, Fang, Eqinfo0 and Eqinfo1 into racks in the
new computer room on Thursday.
- Talked to Ken Ou about ordering an evaluation RAID from Promise. He suggested
talking to Mike Black. Mike has been using Arena RAID enclosures for some time.
He let us borrow one for testing. Installed it on Hotspot and built a wavepool
on Friday afternoon.
- Sick on Monday.
- Resurrected Pluton on Tuesday in order to use the FreeBSD sources on it to
upgrade the disks for Will Prescott's three Athlon machines. These machines
will become the origin servers for the Program site and the Menlo Park site.
Did the upgrades on Wednesday and sent the disks back to Will.
- There was a network problem on Tuesday morning. The 66 net was cut off
from the rest of the campus. Joe from ITS came over and ultimately ended
up rebooting the Catalyst 5000, after which the network returned to normal.
- Chuck moved the phone line for paging modem on Agent86 on Tuesday. Reconnected
the modem.
- Rift died and rebooted itself at about 16:40 on Tuesday.
- Carizo had a keyboard problem on Thursday morning. I was acting like it had
a stuck key. Unplugged the keyboard momentarily to reset it.
- Agent86 had a problem with multiple rsync jobs stepping on each other. This
turned out to be caused by the rsyncs running very slowly, so cron was starting
the next job before the last one was finished. Further investigation showed
that they were running slowly because the network interface on Fang was slow.
This turned out to be caused by the fact that the interface was configured for
full-duplex, but the cisco 2900 switch it was connected to apparently only did
half-duplex. Switching the interface to half-duplex fixed it.
- Rift died mysteriously at about 11:30 on Thursday. The last log showed it rebooting three
times in quick succession. Dissected it to reseat the power supply cables and
also replaced the power cord at about 14:00.
- Fixed a problem with Big Brother paging. The scheme calls for Ehzsouth to be
the primary BBPAGER, with Rift as failover. But the bbwarn settings were
backward. Set Rift to 'bbwarn: FAILOVER' and Ehzsouth to 'bbwarn: TRUE'.
- Put Sangabriel on a shelf in the rack and brought it up Friday morning.
- 1/2 day vacation on Friday.
- Jet crashed at 12:26, 13:58, and 16:34 on Saturday. The console log
showed severe memory errors on board 2, bank 0. Came in and removed
the bank.
- Filled out the travel paperwork from the April 3-4 trip to Menlo Park.
- Started doing configuration on Sqehzeast in preparation for moving the
Program and Quake sites on to it.
- Set up access to Foreshock for Phil Powers in Golden.
- Copied the rtem, tri_db, and tpp directories from Iron to Sangabriel.
- Free-form Fry's odyssey with Bob on Tuesday. Bought the pieces for
the new CIIM map-making machine.
- Started assembling Ciim on Wednesday. The case had a faulty power
supply, and would not turn on. The main board was also flaky. It would
not boot from the floppy, would sometimes hang in POST, logged spurious
disk errors, and numerous other problems.
- Assembled Ciim with the main board from Pluton and a borrowed case.
Ordered a new 2u rackmount case and main board from CalPC. Bob will
take the other board and case back to Fry's on Monday.
- Web team conference call on Friday morning.
- Flint went catatonic from 05:35 to 06:08 on Friday morning. The logs
showed that it ran out of mbufs. There was a crawler going through the
site that sent so many requests that the network stack was swamped.
Rebuilt the kernel with MAXUSERS=256 and NMBCLUSTERS=16384.
- 1/2 day vacation on Friday.
***
- Traveled to Menlo Park for the Earthquake Hazards Program web site
planning meeting.
- Set up the web and mail servers in the new computer room in the
main USGS building.
- Built up a new machine to take over the computation and map-making
for the Community Internet Intensity Maps project. The machine
currently doing it is obsolete, slow, and no longer reliable. The
new machine will speed up processing and improve reliability.