Meeting Notes for 12 September 2007
Present: Liz, Eric, Mihael, Bob, Tom L., Tom J.
Absent: Mike W.
SC2007
We discussed the logistics and who can come. Eric Myers is going to try to come and work with the others as time permits.
Mihael's refactored code
Mihael needs people to test this as thoroughly as possible.
Currently the search pages on IE are messed up. He will fix it and deploy.
Almost every page redirects to a error page. He could add a form on that page for users to fill in.
Bob was able to get the same results on www12 as on production for shower. He needs to see if this extends to larger datasets. The geometry page is now working. www12 The original error Bob noted resulted from Mihael not importing everything that was needed. He has implemented the Geometry passing so the analysis an do the grid thing. Thunderous applause!
Mihael is negotiating on grid sites; for now, he is hoping for a permanent reservation on one of the grid nodes at the UChicago; there are lots of issues related to getting grid certificates.
Currently the Cosmic e-Lab can run with SWIFT on the local machine on www12 port 9080.
We are still working with an alpha version and need testers.
Tom L said he would poke at corners of it.
Bob - Limited dataset a problem; Mihael is working on a mirror of all the data so Bob can do more thorough tests.
Liz: if we get the mirror of the dataset, maybe we can run through jMeter tests.
Once we get a stable e-Lab, we can go on to:
Beta testing by Tom L. and the fellows
This will also test for simultaneous users.
Maybe we can do the JMeter tests.
LIGO Analysis
Mihael and Eric discussed installing Globus on www13 vs. installing ligo tools on Teraport so Eric can test grid execution. On Wiki, installation of ligo tools documented. Eric worried about compatibility with different versions of Linux. He needs a one-time installation of software on grid nodes where we want to run.
Eric has the analysis code running on www13.i2u2.org/tla/
We all went an looked at the seismic data from the huge earthquake in Sumatra.
CMS Analysis
Maybe CMS could use Ligotools. It has Root in it. Tom L. will find out what version of Root CMS uses. Ligo currently uses 4.04. If Dan needs a higher version, it will be more effort because a change in a directory structure.
Backups
Eric contacted MCS to find out what is backed up.
Nfs partition and disk 0 and disk 1
No backups of other machines; home directories are not backed up so it is important to back these up. Eric is rsynching a lot of his work.
www16 and 17 systems are updated. MCS is talking about putting the home areas in NSF.
Login servers are backed up ( login.mcs.anl.gov pool of three machines ). Written over each week, but they have a way to get older stuff back. Backup subdirectories exist.
Mirroring between some of our machines. The database is being mirrored. Eric will document this.
MCS will not run the the demon Eric suggested on the cluster. They use NGIOS Rick is going to see what they can provide us.
Related to this is the issue of documenting how our machines are used and what their status is. This information is in both Current Status and Cluster-Ops. Liz and Eric will discuss off-line if there is a better way to combine this information or avoid duplicating information.
Eric has also been adding Error Recovery information on the wiki.
Issues with www11
www11 is still out of the load-balancing pool, but it is sometimes used by people who are still using the old URL www11. Eric Myers has been looking into Tibi's method of doing the load-balancing and may come up with a more permanent solution that does not depend on the cgi being in Tibi's area. He is documenting what he finds.
Wiki/Forum Issue
(Carry over from last telecon)
Eric is still working on the Wiki for the Fellows. He updated the wiki code for his wikis and has some errors to sort out. He has added the upload capability.
-- Main.LizQuigg - 12 Sep 2007