Minutes of Online Calibration Meeting 11/10/00

Theme: 'Are the Tires Flat'  
         --> NOT DISCUSSED

The meeting consisted of a discussion of database specific 
topics and then a long follow-up to the points discussed 
at the last meeting along with other announcements. No time 
was left for a discussion of DBANA, Consumers, or Valid Sets.

DATABASE:-------------------------------

Replication is essentially finished testing on integration and 
is ready to be cut to production. The online production database 
will have its bad CPU replaced on Wednesday 11/15. In order to 
avoid a blame fest, we will wait until 11/20 to install replication 
on the production machines.

NEWS:-----------------------------------

MAILING LIST: After discussion it was decided to stay with using 
the cdf-daq mailing list because it seemed to reach most of the 
interested parties except for some of the silicon group, and it 
brings in a wider audience.

DATABASE DRIVERS: The code managers are going to drop the OCI 
libraries from the build in development. All groups using oracle 
should make sure they are using the OTL driver. Also, Rick commented 
that they want to turn off the mSQL library build as well for 
Fermilab machines by not setting up mSQL in the setup cdfsoft2 script. 
This is driven primarily by the desire to reduce executable size and 
link times.

CALIBRATION CSL: A new Consumer Server Logger instance has been set 
up for use by calibrations which require all the events for D-mode 
or have very large X-mode events. It is running on b0dap11 and 
can be switched to simply by changing the CSL host in the run 
settings. There are two caveats. First, the data sent to this CSL 
is not logged either to disk or tape. If you want a copy of the 
input data for later testing, you will have to turn on the FileOutput 
in your tcl script. Second, the machine is on a publicly usable and 
resettable machine, so it is somewhat less stable than the other 
CSL instances. Users were encourage to pound on it and give 
feedback to the CSL group.

FROZEN ONLINE VERSIONS: Pressure was applied to make sure that all 
consumers run from Run Control use frozen versions which have been 
copied over to the cdfdaq account. CLC and Calorimeter flashers are 
changing to this now.

INTEGRATION TESTS: The problem code which prevented the switch to 
integration during the last week of the commissioning run was 
anonymously fixed. This was discovered during the tests of replication.

FOLLOW-UP:--------------------------------

Calorimeter Flashers: Philippe was not present, but it was known that 
the desired tables are in development, and he was testing them along 
with A. Kovalev.

QiePed Consumer (J. Cranshaw): The version of QiePed was upgraded and 
built with 3.10.0. It still core dumps at the end.

CLC (A. Safonov): Alexei is switching over to using the error logger 
for messages. During the commissioning run, he was still using the 
development database when taking flasher runs, so there was a discussion 
of whether some of this data needed to be moved into production.

SILICON (J. Slaughter/A. Sill): Silicon has shown they can write to 
production, although most of their commissioning run work was done 
with development. It needs more testing and communication to complete 
the changeover.

Rick asked whether the silcon group had tried using the new capability 
to store and retrieve float arrays. The answer was no, but they were 
starting to work on it. Jack gave a list of the tables which MUST be 
changed over to float arrays.

CEMQIE
CHAQIE
WHAQIE
PEMQIE
PHAQIE
CESQIE
CCRQIE
CPRQIE
PPRQIE
CLACQIE
SIXCHP
SVXDH

As one can see, most of the tables are actually QIE tables; but since 
the silicon group has manpower ready to try it for the silicon tables, 
it was agreed that they should change over first. The changeover will 
affect both consumer and DBANA code.

COT (B. Orejudos): The COT group has been working on 
fixing the problem with the TRACERS which keep it from producing 
reliable timing constants. Bill has also been working with 
P. Tamburello setting up the software for tracking to use these 
constants.

STAGE0 (K. Burkett): Kevin gave a report on experience with Stage0 
during the commissioning run. He was able to take data at roughly 
7-8 Hz which was sufficient. During commissioning and in March no 
dedicated stream is foreseen. Roughly 10^4 hits per superlayer were 
necessary, so it was expected that Stage0 could produce usable drift 
constants in roughly an hour during normal data taking. There are no 
plans to try to change Level3 constants during the run, though.