PPS Pilot Follow-up Meeting Minutes Tue 09 Dec 2008
- Date: Tue 09 Dec 2008
- Agenda: 46152
- Description: pilot of Cream CE: check-point
- Chair: Antonio Retico
Attendance
- PPS: Antonio Retico
- CMS: Apologise
- Alice: Apologise
- CNAF: Daniele Cesini
- PADOVA: Sara Bertocco
- FZK: Absent
- RAL: Derek Ross (observer)
- JRA1/Cream/WMS: Massimo Sgaravatto
- SA3: Alessio Gianelle
Review of action items (tasks)
Status of the subtasks of
TASK:7981 "Set-up and run Cream CE Pilot (Phase2)" (see them in the
PPS tracker ) .
Notes:
The only tasks still in progress are the ones concerning Nagios and SAM. SAM tests for cream still not visible in the portal. To be followed up. Specialised Nagios test
under development at Cern.
All the remaining installation tasks, for CNAF and FZK were closed this week.
Status and results of the pilot service (by VOs and sites)
CMS (absent) had nothing to report. No major activities were carried ot on the pilot in the last two weeks due to other priorities
Updates on layout from Massimo:
- A 'production version' of the Cream CE was installed at FZK. The issues noticed during the installation were due to BUG:44712, known and mentioned among the known issues.
- A Russian site supporting Alice has demanded for help to install the production version of Cream. They had the same issue with BUG:44712 and they needed to be pointed to the workaround. Antonio proposed to attach this site to the BDII used by CMS in order to extend the testbed. Massimo reckons that this would not add value to the test because the submission issues ICE-->old CREAM have been explored already. So this site won't be part of the pilot
- Patricia is in Brazil and we may end-up having a Cream CE there as well
Daniele reports about a new testing activity on the pilot service started by Alice. Some details were sent bu e-mail after the meeting
"The test is meant to validate the cream CE (at cnaf in this case) in order to evaluate the adoption of cream at least at T1s. During these tests Alice will test
CREAM at CNAF only, even if this solution in the future could be used on Italian T2s too.
The duration of the test is not evaluable at the moment, while the number of submitted jobs should be similar to the one used for
production: about 500-1000 job/day."
In order to support this test an additional VOBOX was set-up at CNAF
---++ Status and results of the development (by developers)
Massimo: a tag was released to PPS last week. The WMS submission works well but an issue was observed when submitting with the CLI. Another issue, tracked with
BUG:44454, causes the files in the input sandbox to get corrupted if there is more than one. Surprisingly this was not seen in certification. The fix will be released to the production path.
A new tag is currently under test by Alessio containing optimisations for the proxy delegation during proxy renewal. The release to PPS should happen before the Christmas stop
Open Issues (by VOs, sites, deployment teams)
The usage rate of the service is not exciting. Antonio asks for an estimate for the delivery of ICE to the release track.
Massimo: we are aiming to do it by the end of January. After the release we plan to keep the PPS service at Padova running though for future scalability testing
Massimo points out that the developers would have expected larger participation of PPS sites to the pilot, Padova being for the time being the only active participant.
Antonio: That's true, but it is also true that we don't have from CMS so many reports about heavy utilisation of the pilot and we want to have the minimum installation suitable to serve the users' needs, otherwise we fall back into the old model of PPS from which we want to move away.
Massimo: Alessio is performing scalability testing and having sites more sites outside Padova could add value to these tests
Alessio: the scalability tests performed consists on a large number of jobs submitted with dteam proxy. The jobs are not CPU intensive (5-minutes sleep)
Antonio: that should be ok for some sites supporting dteam (no need for real resources behind). Assuming that we get more PPS sites in what would the preferres layout be (e.g. more sites running a single cream CE or less sites running multiple cream CEs). There cold be an option of PIC getting in the gme, but they cannot grant access to the production queues. As they have described their PPS service to be highly flexible (possibly using virtualisation) they may be able to provide a certain number of cream CEs.
Alessio; Multiple CEs at PIC could be a good case, because they use condor based submission
List of Open bugs and relevant decisions
Recommendations for release and deployment
Decision about termination/extension of the pilot
The decision is made to extend the pilot till the end of January. Next check point to be held on January the 13th at 15.00
AOB