CERN-BARC video conference

CERN-BARC phone conference, 12/7/06

Agenda:

Quattor work - status of wassh re-engineering (Sharma )
Lemon developments (Murthy, Sharma, Salim)
CCTracker (Dinesh, Bill )

Reports:

Murthy:

Memory leak in the case of missing/wrong ssl keys is identified and removed. Performance bottle necks due to nodeid legacy support are removed wherever necessaay

After that full scale tests are conducted again with ~10000 virtual hosts for 72 hours. It is found that the queue turn-around time and accumulation of samples/queue/sec is increasing with time and found to be approaching stable after a very very long time. This is resulting in memory increase with time. It is required to run the test for very long period say 5-6 days and then only conclude on the stability of memory.
Yesterday also a test was conducted for 12 hours and it showed same behaviour. In case the memory is not stable even after some period, ways to be found our to improve database performance.

Email discussions are going on with Miro to resolve the issue.

The OraMon version currently deployed at CERN was not suffering from the memory leak problem.

Dinesh:

CCTracker status:

I've setup a development env here on my PC and deploying and testing the development done at BARC. Once everything works then I will discuss with Bill about the integration plan to deploy it here.

AOB:

Wassh development is currently lower priority until work on sensor-exception has been finished.
German will provide documentation on how to use the XML query API for the CERN-CC wassh site-specific module.
Dennis has circulated a mail explaining how to set up a test environment for sensor-exception. He will add this page to the Lemon Twiki.
In addition, a new Twiki page should be created by Sharma, listing all tests (unit/regression) which have to be passed when releasing a new version of sensor-exception.
Registration for new TWiki users: See this link.

gc 12/07/06