Discussion on PROD Broker network

Wednesday 24 Oct 2012, 16:00 → 18:00 Europe/Amsterdam

   * Update on scheduled upgrade
   * Enable unauthenticated client logging within next configuration rpm
   * Contact major clients
      * SAM
      * OSG (Someone within SAM should know of a contact point)
   * Testing of failover prod broker network capability
      * Test case developed on Ruby (using STOMP gem) works in failover mode
      * According to Perl stomp module documentation this is experimental
         * [AUTH]: Test perl stomp module failover capability
   * Problems identified with current PROD broker network
      * Operations are demanding and need coordination of partners
      * Messages get stuck in brokers and are undelivered
         * [AUTH]: Stress test the broker network with a big amount of messages
         * [AUTH]: Investigate what happens with stuck messages on PROD network
   * In summary:
      * It is not clear to everyone how to resolve problems in the operation of brokers
      * CERN has presented their planning and current work for brokers
         * https://cern.ch/messaging-chep2012 (poster presented at CHEP2012)
         * https://twiki.cern.ch/twiki/bin/view/EMI/MessagingSoftware
      * [ALL]: Investigate possible scenarios (in specific the ones addressed by CERN team) and discuss what seems to be the best option