Meeting Notes for 2 July 2008 Server OS Upgrades Run Levels Database server software Seismic investigation ideas

Meeting Notes for 2 July 2008

Present: Mihael, Eric
Absent: Bob, Liz, Tom L., Dan K, Mike W., Randy, Kris, Joao, Tom J.

The folks from Fermilab were unable to attend due to an all-hands lab meeting. The good news from that meeting is at http://www.nytimes.com/2008/07/03/science/03fermi.html

Server OS Upgrades

Eric noted that the operating system software on our server machines is out of date, and suggested that we schedule an upgrade before we have to have one due to some emergency. Most of our machines, including www18, are running Debian 3.1 (Sarge). Even www16 and www17, which are running a newer Ubuntu version, show "test/unstable" in /etc/debian_release, which is discomforting.

Mihael reported that he's already heard from MCS Systems that they intend to do an overall upgrade, and we would be a part of it. But they know (or should know) that we need to schedule this around ELabs busy times, so the ELabs cluster might be upgraded separately, after MCS goes through their upgrade.

So one question is: when is the best time for such an overall upgrade? There are many workshops in summer, and then teachers go back to school and would like to use the e-Labs in their classrooms. But perhaps the first few weeks of class the usage will be less, and so that would be the best time to aim for? We agreed we need to discuss this with the rest of the group, and keep ahead of the schedule.

Run Levels

Eric suggested that when we do the upgrade that all machines be set to run level 3 for multi-user. Then we can use run level 2 for limited multi-user, such as with NFS turned off and/or certain services turned off by default (so we can turn them on and off for testing).

Database server software

Eric has built and installed MySQL on the database server, data1. Though it is running, it's not yet been properly configured. In particular, it likely won't work yet from other machines. Once that is working he will copy over the Fellows wiki, the Glossary wiki, and the forums databases from Spy Hill so they can be served from www13 (or eventually any other machine we want) from ANL, and take load off Spy Hill.

Eric is documenting the build procedure in the WebBuild page. He also noticed that the postgress we are running was locally built, but Mihael doesn't know who built it or how it was built. Anybody?

We suggest that we consider switching from the built-from-source versions to packaged versions at the time of the OS system upgrade, provided that after the system upgrade there will also be a better mechanism for keeping the system up to date.

Seismic investigation ideas

After the main business was completed we talked a bit about details of implementing more transformations in Bluestone, and what kind of e-Lab activities this could lead to. In particular, a cross-correlation of two or more detectors could give difference in arrival times of seismic signals, which could be used to determine the direction to an earthquake. This is in contrast to how USGS does it, using triangulation. The time difference between P and S waves may give distance to the event. If this kind of analysis works one might be able to create a real-time or near real-time display of earthquakes or other disturbances around the LIGO sites, perhaps even individual trees being cut down near Livingston, or snow plows driving around after a snowstorm near Hanford. Getting there will take a lot more work, but we enjoyed getting excited about the prospects before going back to work on the dull details.

We also talked a little bit about the USGS Earthquake Notification System (ENS), at http://earthquake.usgs.gov, where you can subscribe to e-mail notifications based on region, magnitude, and time of day. This would be useful for students to get a feel for the frequency of earthquakes. Eric gets these, and he has a perl script which extracts the main parameters and then sends them to his cell phone via text message. Note also: the USGS has another machine for development work, with almost the same name, but it is "earthquakeS" not just "earthquake". Eric got on that once by accident and didn't figure out the problem until he corresponded with one of their developers. At least we only have the problem of people mistaking www13 for www18.

-- Main.EricMyers - 02 Jul 2008