Go to the Indico Home Page Oldest event Older event Up to category Newer event Newest event Printable version
  • Event
  • Lecture
    • Switch to the management area of this event
    Login
    JRA1-UK

    Weekly Meeting

    by Steve Fisher

    → Europe/Zurich
    CR13

    CR13
    Description
    <html> <head> <meta name="generator" content= "HTML Tidy for Linux/x86 (vers 1st March 2002), see www.w3.org"> <title></title> </head> <body>
    • Reports on Meetings
      • EGEE in Pisa - Steve F etc.
    • Future Events
      • CHEP'06 in Mumbai- Progress on paper - Steve F
      • HPDC in Paris
    • Actions
      • Write test specification. - John - 1/8/2005 Will be updated when the tests are stable
      • Track Dave K's progress to check R-GMA publisher in L&B test. - Antony - 13/10/2005
      • Publish user guides and APIs on web for 1.4 and 1.5 - Antony - 20/10/2005
    • LCG2_4_0/2_6_0 - Antony
    • APEL and SD - Steve F
    • LCG support - Steve F
    • gLite 1.5 and LCG2_7_0 or ????
    • Bugs - (prioritisation)
      • #7024 Additions to the manuals - John
        Java done and C++ almost done, C half done and Python to do. The is now more work for 1.5 so this bug remains open but is transferred to John.
      • #7226 rgma client configuration succeeds, but a client test fails - Martin.
        It was decided to follow the java trustmanager style for priority and if nothing is set it crashes.
      • #7276 Bad error message for expired proxy - Martin
        See 7226
      • #7399 API errors with expired proxy - Martin
        See 7226
      • #8099 Archivers are inconsistent - Martin
      • #8356 Test the service discovery API - John
      • #9375 glite-rgma-server-installer does not create /var/log/glite/rgma-server/rgma-server.log - Alastair
        This may require 2 bugs for YAIM and Dep Modules
      • #9395 Registry API causes deadlock - Rob.
        Low priority - is this still true?
      • #10611 SD does not try the second option - Akos.
      • #12234 Service Test/ODP/testDeclareStaticTableInvalidPredicateColumnNames etc - Steve H.
        Low priority - the ODP needs reworking.
      • #12236 Service test/ODP/testAbortClosed: Does not throw UnknownResourceException - Steve H.
        Low priority - the ODP needs reworking.
      • #12238 Service test/PPP/testCreateNegativeQueryType etc fail - Steve H.
        These checks are not necessary. The calls are made by our own code in a very restricted way. Ultimately, refactoring will remove this problem altogether (by replacing the "int" parameter with two booleans).
      • #12243 Service test/PPP/testGetHRPNotDeclared etc fail - Steve H.
        I could fix this but it would be a kludge, as there is no per-table LRP or HRP yet, so whatever table you specify (whether it exists or not) will return the same value. I guess this will only affect the user if his code currently works OK specifying an undeclared table, but will fail in a subsequent release when the proper functionality is included.
      • #12245 Service test/RP/testRegisterProducerTableZeroHRP etc fail - Steve H.
        I guess this bug should remain unfixed and the test continue to fail until we've removed the old system APIs completely, at which point fixing the test won't break backwards compatibility.
      • #12249 static producers are registered using the same flags code as history producers - Steve H.
        If a new static producer will ever talk to an old Registry, this can't yet be fixed as the Registry will report an invalid flag. Otherwise it should (I think) be simple. We should be able to assume that registry is updated first.
      • #12253 SFS Test Consumer/Join sometimes fails - Martin.
        Just transferred from Steve H to Martin.
      • #12255 SFS Consumer/MultipleProducers - FAIL (6 & 7) - Steve H.
        Needs further investigation. May result from the entropy problem.
      • #12256 SFS test Consumer/MediatorWarnings - FAIL(1 2 & 3) - Steve H.
        Martin is right that giving people a CONTINUOUS/MEMORY producer when they ask for LATEST/MEMORY or HISTORY/MEMORY is plain wrong. In fact, we've told people in the User Guide for the last year or so that LATEST/MEMORY, HISTORY/MEMORY and LATEST+HISTORY are not supported and would throw an exception if you asked for them (so mapping LATEST/MEMORY to an old LatestProducer is also wrong). Most of the APIs have enforced this up to 1.4 and I think we should continue to+do so in the server, now that we've removed these checks from the APIs to get consistent error messages from the server.
      • #12258 SFS Test OnDemandProducer/SocketProtocol - FAIL - Steve H
      • #12260 SFS Timeouts/RetentionPeriods - FAIL - Steve H.
        The LRPs aren't yet stored in the tuple, so there is no way that SecondaryProducers can honour them. For PrimaryProducers, we could fairly easily alter the current situation so that Latest producers use the LRP as their minRetentionPeriod, while DBProducers still use the yHRP.
      • #12518 overview: DB connection pooling required - ???
      • #12604 C++ API doesn't handle Tomcat shutting a secure connection - Alastair
      • #12645 NULL mediator warning on empty result set in C API - John
    • gLite 1.5
      • APIS - John
        • ParameterChecks. "Will be done this week"
        • Expand ServiceCalls tests to check a bigger range of result sets including: Empty results sets, Result sets with each column type, Result sets with empty table and/or column names in metadata "Will be done this week"
        • Need to automate check of insecure connections.
      • Function
        • Still to be defined. Low priority - John.
      • Distributed Full System
        • Probably not going to get around to these for 1.5. Lowest priority.
      • Tools - Alastair
        • Command line - Bug found.
        • rgma-client-check - Failed in CL tool but should be fixed.
    • AOB
    </body> </html>
    actionlist
    minutes
    transparencies