PPS Pilot Follow-up Meeting Minutes Fri 24 Apr 2009

  • Date: Fri 24 Apr 2009
  • Agenda: 57611
  • Description: Pilot of glexec/SCAS: check-point
  • Chair: Antonio Retico

Attendance

  • PPS: Antonio Retico
  • CERN: Gianni Pucciani (SA3)
  • CMS: -
  • Atlas: Jose Caballero, Massimo Lamanna
  • Alice: -
  • LHCb: Apologies
  • Nikhef: (Development) - Dennis Van Dok, Mischa Salle
  • IN2P3: Pierre Girard
  • FZK: Angela Poschlad
  • Lancaster: Peter Love

Review of action items (tasks)

SA1/SA3 tasks

Status of the subtasks of TASK:8986 (see them in the PPS tracker ) .

  • TASK:9073 (Service Management at FZK):
    • Angela: nothing new to report in the past three weeks. Are we really supposed to publish our CE as 'production'?
    • Antonio: This is for sure envisaged by LHCb and Atlas. An issue was recently found in PPS deployment test where apparently the compatibility between glexec and CREAM is broken by this version (PATCH:2892). As the conditions in PPS are currently exactly the one that this software would find in production, we want to have this issue fully understood and solved before proceeding. (from the follow-up of the meeting the recommendation for Angela was to wait for glexec PATCH:2973 to be available and to try it before publishing the resources in production)
    • Massimo: actually during the past week we already experienced issues as we discovered that GANGA was using CREAM and of course filing
    • Antonio: in the current configuration (GlueCEStateStatus="Special") CREAM cannot be matched accidentally.Last week we registered a configuration issue at CERN and at a Russian site. These were followed up and fixed. Our concern are not due to the WMS-CREAM patch, which we know is not working, but to the effect that this incompatibility with glexec may have on the CREAM nodes in production, which are heavily used by Alice.

  • TASK:9333 (Installation at IN2P3)
    • Pierre: I re-started working on the installation few days ago and I found that I was using an old version of the lcmaps library. This is probably due to the fact that I didn't use an "official" relocatable WN but created my own one with glexec. As a tarball is now available (by Ricardo Mendez) I would porpose to re-do the installation in the right way. Unfortunately I am also the reviewer of a JRA1 deliverable and this will keep me busy the next week, so I cannot promise to dedicate much time to this installation. With respect to the issues with the configuration issues reported the last time (configuration of the ld.so.conf needed) I negotiated with the people responsible of the WNs and I obtained the base configuration to be changed (only for this time).
    • Gianni: We discovered after the delivery of PATCH:2892 (rejected from PPS) that actually the glite-security-lcmaps rpms were missing, which is likely to have caused this issue. The version of lcmaps used in certification was 1.3.10-2 . A new tarball will be created out of the replacing patch.
    • Mischa: the new PATCH:2973 will contain the correct dependencies as well as the fix for BUG:49493 (issues observed with CREAM which are now understood). It could be ready by Monday, but as Oscar is back the same day maybe that he wants to have a look before to double check.
    • Antonio: So the expected timeline for the set-up of IN2P3 could be
      • Tuesday 28/4 : PATCH:2973 released
      • Wednesday 29/4 : tarball created and available to Pierre
      • Installation at IN2P3 carried out afterward (should be finished at some point between the 4th and 8th of May)
        There is no strict constraint by LHCb on this date. Actually LHCb reported that they are happy with the integration of the error codes they did. For the next future they plan to integrate in their production new sites as soon as they will be available.
    • Massimo: Atlas is preparing to gether with CMS for the STEP09 exercise and it would be very important to have done experience with the use of glexec in real production analysis jobs by that time (first half of June). This date is not very far.
    • Pierre: Question. is the relocatable distribution going to include also the myproxy logon client needed from Atlas?
    • Gianni (answered off-line after checking): yes, the myproxy logon client will be supplied by the WN package

Antonio welcomes Peter Love (Lancaster, UK) as new participant of the glexec/SCAS pre-production exercise.

  • Peter: In the context of Atlas analysis our site (~600 job slots) is using pilot jobs playing around with our solution for STEP09. We currently submit using production credential and we need to start using glexec to switch credentials.
  • Antonio: this is perfectly compatible with what we are doing here and Lancaster is welcome to join. I'll get in touch with you for the few formal registration steps to do. Information about this activity including installation and configuration instructions for the sites is available at PpsPilotSCAS . Be aware however that there are known issues with the existing patch and therefore it is strongly recommended to start after PATCH:2973 will be released next Tuesday
  • Peter: what is the recommended deployment scenario? e.g. different CE,separate WNs?
  • Antonio: the glexec should be deployed on the production WNs. Then the way these are accessed is up to you. e.g. FZK has provided a second CE, NIKHEF has published everything through the production CE. I think we have passed the phase in which we needed buffer services so a deployment in production can be done if you feel like. If not, feel free to install a second CE
  • Massimo: actually from the Atlas perspective installing a second CE can be preferable (just to make sure to have a fall-back)
  • Antonio: you can ask for support on the installation both to Pierre (IN2P3) and Angela (FZK). I think that Angela's set-up is more similar to the one you are going to make.

other tasks


Notes:

Status and results of the pilot service (by VOs and sites)

Atlas

nothing to report for this period

LHCb

nothing to report for this period

Status and results of the development and certification

Gianni: rejection of PATCH:2892 as already mentioned

Open Issues (by VOs, sites, deployment teams)

none

Decision about termination/extension of the pilot

Nothing was discussed, An extension will be needed (currently finishing by 30th April). Likely shifting to end of May

AOB

Pierre: is glexec supported on SLC5 and on 64bit?
  • Gianni: we did not received a 64bit glexec in certification yet.
  • Dennis: there is a version in Etics. I ma not aware of any problems. it should be working
  • Antonio: to be followed up at the EMT

Next events

The next check-point meeting will be held on May the 12th (Tuesday) at 16:00.


Edit | Attach | Watch | Print version | History: r5 < r4 < r3 < r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r5 - 2009-04-27 - AntonioRetico
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LCG All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback