ALICE- LCG Task Force Meeting 28-Aug-2008


People Present

Room: Betev, L; Mendez, P.; Schutz, Y.; Carminati, F.; Koulouzis, S.; Saiz, P.;Meoni, M.; Furano, F.; Padee, A.;

Phone: Kowalski, M. ; Schiaua, C.; Barbet, J.M.;

Minutes taken by: Saiz, P

__________________________________________________________________
Status of CREAM (Mendez, P.)   (CERN)
******************************************************************
The slides are on the agenda page. The CREAM CE is stable. 

This week, the CREAM CE in FZK points to the ALICE queue. They have to publish the new queue  

We need to define the OutputSandbox only if we want to get the output of the JobAgent.

Lat: Claudiu, do you want to run the CREAM in NIHAM?
Cla: I've already ask the admin. 


Pat: We are the only VO testing CREAM
Fed: If we get into trouble, can we expect any support?
Pat: Yes, of course. We are using CREAM in the way it is supposed to be. No backdoors,


Questions:

Lat: Jean-Michel, this will simplify your life
JM: Sure, I will have a look

__________________________________________________________________
Job Requirements and automatic data transfers  Spyridon Koulouzis
******************************************************************

The slides are on the web. I have implemented a retrial mechanism for the FTD.
When you mirror a file you can specify the number of times it should be tried before giving up


Lat: How does this work with the global redirector?
Fab: I have nothing against this.  With the GR, when a site is subscribed, you only ask for the prestage of the file

I have also created a new command that gives the reason why jobs still wait in the queue. It analyzes the requirements
of a job, and specifies the ones that can't be fulfilled. 

Lat; Please, change the name. It's difficult to remember. Could you put it as 'jobListMatch -verbose'?
Spy: Ok

The command also has an option to fix the missing requirements. For the time being, this can only work with jobs that
have data in several sites. 'Fixing' the req

Fed: An indiscriminate fix is a bit too dangerous. There should be limits
Spy: Ok, we can put threshold in the % and total amount of space


Questions:

Fed: That's very nice, because people will know what's going on


Lat: How long are you going to stay here?
Spy: I'm leaving tomorrow.


_____________________________________________________________________

First Physics Production cycles  (Betev, L.)
*********************************************************************

We are running a complex production for first physics. We have to redo a third of the jobs.
The users are checking the configuration. After a week, we will come back to the normal situation


We are testing a couple of SE, Kurchatov and Madrid. Both sites are DPM. With Madrid there are no problems


We have some problems with Gridka. This is really bad, because part of the raw data is there. 


The beam will come in the 10th, and there will be huge pressure to have the data

JM: We have a high rate transfer from the SE
Lat: As Claudiu said, someone is reading the data