Go to the Indico Home Page Oldest event Older event Up to category Newer event Newest event Printable version
  • Compact style
  • Indico style
  • Indico style - inline minutes
  • Indico style - numbered
  • Indico style - numbered + minutes
  • Indico Weeks View
    • Switch to the management area of this event
    Login

    PASTE meeting

    → Europe/Zurich
    2-R-014 (CERN)

    2-R-014

    CERN
    Andrei Tsaregorodtsev
    Description
    CERN, bat 2 R-014 To join the telephone conference, click here: https://audioconf.cern.ch/call/0165202 To dial in to the conference: a. Dial +41 22 7676000 b. Enter access code 0165202
    Hide

    PASTE Meeting, 8th February 2011, CERN

    Chair:  A.Tsaregorodtsev
    Minutes:  F.Stagni

    CERN: F.Stagni, R.Santinelli, J.Closier, E.Lanciotti, V.Romanovsky, K.Ciba, S.Reuser, Z.Mathe, R.Santana, M.Rashid, M.Ubeda
    Phone: A.Tsaregorodtsev, A.Casajus


    We didn't go through the Savannah tasks today. We discussed the following points:

    # Production issues

    - MC production

    Vladimir: the main problem is replication transformations: all "jobs" for files distribution are still waiting. This is due to a bug introduced in the Request Server with the last patches (a missing method).

    - Running at high rates:

    Spikes in the number of jobs caused load problems. Ricardo duplicated the JobStateUpdate servers and it worked perfectly. The rules of the IP tables have been modified. A better solution is to create several URLs to balance the load. Such solution will be tested in the certification system. This solution does not require any new release. It is a pure configuration issue.

    In the last week we've run at a rate of 25k concurrent jobs (which is a new record) and about 50k jobs per day, running 1.5 jobs per pilots.

    - Efficient use of sites:

    Andrei: we have arrived to a point in time when we can move to a system where we don't rely anymore on rankings and WMS, but using direct submission, so that we can control it better how many pilots are sent. We can get more resources that what we get now. Ricardo and Roberto don't think we'll get more resources.
    We can put more sites in the direct sumbission mode in the next weeks (right now CNAF T2 is running with direct submission), and carefully watch it. The new release provides updated machinery to comply on it.


    # (LHCb)DIRAC Release Update

    - v5r11 patches

    The last DIRAC patch was lacking some bits. We need one more (DIRAC only) patch to fix the problem with the Request system. This patch will be only for the server side. Also the LHCbDIRAC part needs a fix on RSS should be deployed.

    - v5r12 release status and certification tests

    The DIRAC v5r12 is ready but it is not used at the moment by LHCb. DIRAC v5r13 (or v6r0?) will be next release, to be certified together with LHCbDIRAC v6r0.
    A new pre-release can be pre-prepared.

    On Tuesday the 22nd all the machines will be shut down.

    Plan:
    - Friday evening we send a mail that we'll stop the system (stop accepting user jobs).
    - Weekend will be used for draining the system.
    - Monday we update the system and start checking.
    - Tuesday morning the machine will be rebooted.
    - The system is still stopped for the users on tuesday afternoon, checks can be made by us.
    - Wednesday morning we re-open the system for the users.

    Joel will send the announcements.


    - Certification procedure

    Ricardo: we should formalize the way we report the problems to the developers. Savannah could be used.

    A migration plan should be drafted, but for the moment we don't have the time to make tools for smooth migrations (e.g. for MySQL schema changes). Joel will collect such information. In this migration plan we should also write down which services/agents have to be started.

    In https://lhcb-cert-dirac.cern.ch/doc/_build/html/ there is a draft regarding certification procedure (no updates in the last week).

    # Data Management issues

    - Integrity checking

    Being tested on volhcb12. Tests are made with a SARA dump.

    All the other discussions could be made on the thursday meeting that is dedicated (mostly) for data management.


    # AOB

    Ricardo is preparing a document on how to make use of branches and tags for release procedures. This document has been discussed with Hubert and Marco Cl. This will be circulated soon. Every developer should have a look.

    DIRAC week: when? This will have to be discussed because it seems that the week of the 21st clashes with many other things. Ricardo is proposing to post-pone it of 2 weeks. Maybe a doodle could be opened.

    There are minutes attached to this event. Show them.
      • 09:30 → 10:00
        Production issues 30m
        - Running at high rates - Efficient use of sites
      • 10:00 → 10:20
        DIRAC Release Update 20m
        - v5r11 patches - v5r12, v5r13 DIRAC releases - v6r0 release status o planing o procedures for migration o Documentation policy - Certification procedure
      • 10:20 → 10:40
        Data Management issues 20m 2-R-014

        2-R-014

        CERN
        - Storage Usage Accounting - Data archival
      • 10:40 → 10:50
        AOB 10m 2-R-014

        2-R-014

        CERN
        - DIRAC week
      • 10:50 → 10:55
        Ongoing actions (as reminder) 5m 2-R-014

        2-R-014

        CERN
        The task list is available via the Savannah DIRAC group ( Task List ). If you are not already registered please do so at the Savannah page. In order to streamline the time spent during PASTE on the actions list only tasks in the status 'ToBeDiscussed' will be discussed there. In case some actions fall by the wayside they will automatically be updated to this status. Therefore the action assignees should take care to update this. For ongoing actions it is advantageous to report any progress in the Savannah tasks each week.