Uploaded image for project: 'AMQ Broker'
  1. AMQ Broker
  2. ENTMQBR-7535

[Doc] Document HA configuration for Federation

XMLWordPrintable

    • False
    • None
    • False
    • Hide

      +underlined text+1. Set up a shared-store ha pair of brokers on two hosts to simulate one datacenter

      2. Setup a second shared-store ha pair on two other hosts to simulate the second datacenter

      3. configure federation for 20 queues in each direction (examples attached)

      4. Start up 1 consumer for each queue on either end of the federated cluster

      5. Start up all 4 brokers and check the console for each live node. We see 40 consumers on each live node - 20 local and 20 via federation.

      6. Stop the master on one end and wait a few moments. Check the backup and verify that consumers are still shown from the opposite federated node

      7. Stop the backup on the same end, wait a few moments with both brokers down, then start the master back up.

      8. Check the console after a few moments - only local consumers and connections are shown

      Show
      +underlined text+1. Set up a shared-store ha pair of brokers on two hosts to simulate one datacenter 2. Setup a second shared-store ha pair on two other hosts to simulate the second datacenter 3. configure federation for 20 queues in each direction (examples attached) 4. Start up 1 consumer for each queue on either end of the federated cluster 5. Start up all 4 brokers and check the console for each live node. We see 40 consumers on each live node - 20 local and 20 via federation. 6. Stop the master on one end and wait a few moments. Check the backup and verify that consumers are still shown from the opposite federated node 7. Stop the backup on the same end, wait a few moments with both brokers down, then start the master back up. 8. Check the console after a few moments - only local consumers and connections are shown
    • Critical

      Current documentation does not describe the requirements for making federation work with HA endpoints. Setting the <ha> element to true and adding the connectors for the primary and backup brokers doesn't make HA work as expected. We also need to make sure that reconnect-attempts and other attributes are configured appropriately for failover behavior of core API clients. Something like this seems to work:

            <federations xmlns="urn:activemq:core">
         <federation name="rr-federation" user="admin" password="admin">

            <upstream name="rr-upstream">
               <ha>true</ha>
               <circuit-breaker-timeout>5000</circuit-breaker-timeout>
               <reconnect-attempts>3</reconnect-attempts>
               <call-failover-timeout>-1</call-failover-timeout>
               <static-connectors>
                  <connector-ref>upstream-primary</connector-ref>
                  <connector-ref>upstream-secondary</connector-ref>
               </static-connectors>
               <policy ref="rr-queue-policy"/>
            </upstream>

              jcliffor@redhat.com John Clifford
              rhn-support-dhawkins Duane Hawkins
              Votes:
              1 Vote for this issue
              Watchers:
              4 Start watching this issue

                Created:
                Updated: