Deployment team

Tuesday 7 Aug 2007, 11:00 → 12:00 Europe/Zurich

EVO - GridPP Deployment team meeting

Jeremy Coles

- This is the weekly DTEAM meeting - The intention is to run the meeting in EVO: http://evo.caltech.edu/evoGate/. Join the meeting in the GridPP Community area - The backup plan is to use VRVS - Sunny virtual room

Minutes DTEAM_minutes-2007-08-07.txt

11:00 → 11:20 Experiment problems/issues

Review of weekly issues by experiment/VO - LHCb - CMS - ATLAS -- SL tests show particular problems (full queues or SD?) for UCL-HEP (0%); BHAM-HEP (0%); RAL-Tier-1 (2%). - Other

11:20 → 11:25 Areas in need of progression

- Glue-schema From 25th July: "Dear ROC Managers and Site admins, so far we have received only 2 use cases for the new Glue Schema 2.0 please let us now if you have more. I remind you that a draft document for use cases can be found in the OGF gridforge repository. http://forge.gridforum.org/sf/go/doc14621 Please send your suggestions to: egee-tcg-site-rep@cern.ch thanks..." What if any feedback has been provided? - Signals from killed jobs ...according to the documentation qdel sends a SIGTERM and then a SIGKILL to the jobs. The delay between the signals is configurable by the system administrator either at server level or at queue level. SIGKILL is only sent if the job hasn't exited by itself in the meantime. - ROC storage security challenge -- From 10th July "... 1) Send an email, below a draft that you can comment upon, to sites (tb-support and gridpp-storage) 2) Run the actual jobs (next week) 3) Wait 1 or 2 weeks (I would wait for longer but then it becomes a nightmare for me) 4) Send the tickets." What is the status?

11:25 → 11:40 ROC update

- TCG voting status: just 5 responses so far! - Tickets outstanding - future management of tickets - news from ROC managers meeting EGEE ops meeting items of interest: *************************************** -- Original statement: The EGEE PMB recently made the decision that the gLite CE is not of production quality and will be depricated (in the near future). Therefore sites should put no effort into this service. Development effort will go into the CREAM CE and the LCG CE to make them production ready on SL4 with VDT 1.6. AMMENDMENT: There has been no decision by the EGEE PMB regarding the future of the CEs in the EGEE grid. The PMB had requested that the CREAM and gLite CEs were tested until the end of July in order that a decision can be made as to which CE will be developed further for production. This testing is now completed and the decision is pending. In the meantime, it should be noted that the version of the gLite CE in the repositories is not suitable for deployment in production (as stated in the release notes). -- We have a number of external dependencies which can be satisfied by using the 'DAG' repository which is a 3rd party repository containing lots of extras for SL4/RHEL4. It is even enabled by default on SLC4. The question is - is it acceptable to require administrators to point their installations to this? -- SA3 is now maintaining a simple report of node type against readiness for SL4. This has been requested in the past. https://twiki.cern.ch/twiki/bin/view/EGEE/Glite31NodeTracker -- Update on YAIM config tool * Data schema changes on the CIC portal side are being carried on, and estimated deadline for this to be finished is 1-2 weeks. * In parallel, implied changes in the portal code (VO registration and VO ID card update forms) will be implemented within the same deadline. * These changes have been discussed and agreed, they also answer requirements from the VO managers group. See my mail on this topic for more details * AS the issues above are the most important, I've not started working with Dimitar on the integration work of the web part of the tool. This should start next week. * All in one, estimated deadline for the whole stuff to be operational and in production is in 2-3 weeks.

11:40 → 11:45 Supported VOs

- Status of regional VOs - g-Eclipse (what else needs to be provided?) - minos and supernemo

11:45 → 11:55 Site availability status

- SAM test results today are excellent! 24hr average across all sites is 97%. - Yesterday at the PMB we looked at the main monthly summary of results http://www.gridpp.ac.uk/wiki/SAM_availability:_Monthly_summary_table and then at the detailed plots http://www.gridpp.ac.uk/wiki/SAM_availability:_May-July_2007. Only two sites had not offered comments (sysadmins away?). - While not a perfect measure, the SAM tests are being adopted as a GridPP measure of site performance together with the KSI2K delivered. They will play a part in the algorithm used to allocate GridPP3 hardware. There are times when a site is exposing itself to avoidable performance degradation in the interests of GridPP sites as a whole (such as first deployment of a new release). We need a mechanism to agree when not to count SAM test results for such sites.

11:55 → 12:00 AOB

- Actions - Swamped CEs (see David's TB-SUPPORT mail)