- Jet had a problem auto-booting after Phil upgraded the console firmware.
The fix was to set 'diag-switch' to 'false' at the Openboot prompt.
- Assisted with setting up station AGA on port 4 on the CVWD terminal
server.
- Set up QDDS, CNSSM, and the new Simpson map on Ehzsouth.
- Put one of the new 9GB disks on Ridge. It was necessary to install a
patch to the scsi drivers to make this work.
- Wrote up an article about the Squid server and the testing that was done
on it.
- Tried to talk to DWP over the microwave link. It was not working. After
talking to their network person, they fixed the problem on their router,
but then we still were unable to talk to our dataloggers. The problem was
in their firewall, and they are unable to fix it at this time, due to a 'no
changes until after Y2K' rule.
- A M3.9 event in Orange County generated a 32 hits/second traffic spike on
the web server. The Simpson map on scec was brought down immediately. As
a result, we changed our links to put the new Simpson map on Ehzsouth into
production. The Squid weathered the traffic spike easily.
- Did some more stress testing and performance tuning on the Squid server.
Increased the maximum file descriptors in the kernel, recompiled Squid, and
set the setwork interface to full duplex.
- Did the VMS 7.2 upgrade on a copy of the Bigone system disk. After the
upgrade, the system hung during reboot. This turned out to be a problem
with the queue manager. As a result of this, we were unable to complete
the upgrade at this time.
- The disk in slot B-09 on the Jet HA RAID failed on Monday night.
Replaced the disk with a cold spare and called n-Stor to get a replacement.
- Made some adjustments in the Squid configuration to allow it to cache
things from the /gifs and /logo directories for a longer time, as these are
static objects.
- Checked up on the Y2K patch status of the Solaris systems. Sun had a new
patch cluster available, dated 12/9. The new cluster had two more patches
than the old one, and five of the nine original patches had new versions.
Reinstalled the whole cluster on all systems.
- Re-partitioned the disk on Stannum to make a bigger /ver area so that the
new Y2k patch cluster for Solaris could be installed.
- Talked to Jim Fisher in Reston about the procedure for getting a name of
the form xxx.usgs.gov/ for our web server. The upshot is that they don't
want to do it unless we come up with a very convincing argument. As a
stopgap, had Charlene Fischer in Menlo Park add the name
pasadena.wr.usgs.gov for our web site. This is so that we can have a name
without the hyphen character, which seemed to confuse people so much.
- Assisted Katrin with benchmark testing on the new SCEC data center web
server. They are running an Apache server with a Squid cache on the same
machine.
- Did some more testing and performance tuning on the USGS Squid cache
server. After tuning, it was able to handle up to 367 requests/sec without
slowing down. This is about a factor of 5 larger than the peak traffic
experienced after Hector Mine.
- Started assembling the new E450 server for the Shake Map group.
- Set up the disk mountings for Hotspot, which is the new E450 for Trinet
development.
- Made an alias for 'webmaster' on usgs-squid. This is so that mail sent
to webmaster@pasadena.wr.usgs.gov will be properly forwarded to Lisa Wald.
- Added some updates to the Simpson map programs per Bob Simpson.
***
- The Squid server proved itself under fire after an event on December 7th.
It was responsive when other servers were dead.
- Reconfigured the Squid server to handle over 350 hits/sec in testing.