Minutes SLAC-ALTAS TDAQ Meeting 07-Dec-2006
-------------------------------------------

Present: Su Dong, Rainer, Stephen, Sarah, Amedeo


Stephen:

Put together tests.  All but PT1.  One problem showed up, a warning
printed to stdout with a timestamp.  Will get that changed to stderr,
and drop the timestamp so it can be diff'ed more easily.

Request to monitor CPU times for all modules.  RTT looks easy to use.

Stored this with MT+PT releases.  Need to get it where RTT is run.
RTT gets run at UCL.

Not yet running a combined test, just one slice at a time.


Sarah:

1000 nodes now, 930 nodes alive.  Even in the next day another 200 or
so.  May meet the goal of 1200.  Useful for testing large partitions.
Also helps parallel work going on.

Only problem right now with reading the geometry database.  A remnant
of the CORAL interaction.

Can't use geometry with or without the proxy.  There is a patch that
has to be used.

Timing test with the proxy.  Comparing running with the proxy with
running from COOL.

Affected by concurrent ~50 rsync process using 100% CPU.

The machines the proxies are being run on are picked randomly.  This
makes the test much less controlled.

There is some indication that Serge saw timeouts at ~250 nodes that he
didn't see with the proxy.  So that may already be running against it.
Will get the machines at 5am next morning for test with DbStresser.
Previous timing results are very odd.  Suspect that this is from
external node on the machines.