Present: Tibi, Liz, Mike Jr. (Mihael), Mike W.

Accessing www13 with Dreamweaver

Kris W. and Tom L.'s work on www13 using Dreamweaver is proceeding.

Routing of users from www.i2u2.org

Liz reported making more progress on new pages for www13.i2u2.org/ that will provide background on the project.

Jmeter and Load-balancing

Bob Peterson worked on one JMeter test, but was unable to save it. Tibi reported working for 1.5 days on investigating load-balancing. Method 1: One was the apache-recommended one in which the first request takes the user to an explicit machine (e.g., www13) and stay there for the whole session. With cgi script - it does not help a lot to get statistics. Method 2: always see www.i2u2.org ( like Google).

Mike W's suggestion: Get all the machines involved in the load-balancing. Check that the routing of all users is o.k. so that most users are getting load-shared. Make sure that the cloning of the machines is the same. Don’t spend any more time trying to achieve perfect load-balancing. A simple random or round-robin is fine. Recalibrate task.

Analysis Code

Tom Jordan reported that he finished a draft of the type of script that can execute a Cosmic transform from the command line. It is in the (CI) cvs in /src/perl/ and in his area on www13. This can be used to try to analyze what code is eating up the cpu. He needs to test the changes he is making to the plot that did not get into the rollout. He had some problems with www13 giving the Change your parameters error. Tibi pointed out that these errors were caused by there being no actual data associated with the results of the searches. Tom would need to upload that. Tom encountered another problem which Tibi fixed by adding a missing piece.

Rollout

Mike Jr. did not do a rollout last Friday because there were no changes. He hopes to do one on Friday to include Tom J.'s changes to the plots. He can use www16 if www17 is broken to test the rollout.

State of www13

Liz indicated concern we did not have the latest versions of code on www13 so it was hard to test new functionality without the confidence that everything was working before the test versions were added. She suggested that after Tom J. finished his tests that we should do a rollout of the current production code. Tibi will do this next week.

--+++Mike Jr.'s progress on update of code Mike Jr. is working on saving plots. He moved the conversion of the svg file to pngs into the actual workflow. He also worked on showing a progress bar during analysis.

Dead Machines

Mike. Jr. tried to communicate with Systems regarding hardware problems; tried to use original ticket. Mike W. suggested he copy everybody on any email to Systems. Mike W. will send one message now. Mike Jr. thinks that there is a heat problem. 14 and 15 are working according to Tibi; 16, and 17 are not working. Does 17 have a hard disk error or is it some kind of software issue that Rick was able to fix on another machine,. Get Rick more involved. Systems should troubleshoot at kernel and below then it is up to us for checking that everything is working.

Action Items:

For Tibi, Bob and Mihael:(left over from last week) going to put a two week deadline to: May 2nd to get some jmeter tests running with best load-balancing scheme and explore how to test that they are working.

Mike Jr.. - will communicate with systems about hardware problems.

Mike Jr. - will look into why the annotation of variables for transformations is failing when Mike W. thinks the metadat contains the information.

Liz - will continue work with higher level pages for i2u2

Tom J. - finish testing of plot and put it into CVS

Mike W. - working on ELabs proposal

Tibi - load-balancing task as redefined above.

-- Main.LizQuigg - 26 Apr 2007