Too many LISTENER processes (11gR2 Grid Infrastructure (2 nodes))

Blues Breaker
    Greetings -

    I get an alert that says that my LISTENER is down (EM Alert Details). It suggests a corrective action
    to stop the current running LISTENER and start it again using the listener parameter (listener.ora) from
    the database (RDBMS) home. There are two (2) LISTENER processes running (ps -ef | grep LISTENER) .
    One process is running from the grid home and the other from the database home however only
    the one running from the grid home has a listener parameter file (listener.ora) there are no others.
    To confuse me even further the documentation says that I should not set the 'local_listener' parameter
    and that the Oracle Agent will maintain this setting.

    Any suggestions ?


    Thanks in advance.

    Brian
      • 2. Re: Too many LISTENER processes (11gR2 Grid Infrastructure (2 nodes))
        Mufalani
        Hello Brian,

        Are you using SCAN (Single Client Access Name) feature? If yes, there are LISTENER_SCANs running on your machines.

        Check with command from your $GRID_HOME

        srvctl status scan_listener

        I think this is a great doc for you --> 11gR2 Grid Infrastructure Single Client Access Name (SCAN) Explained [ID 887522.1]

        Best Regards,

        Rodrigo Mufalani
        http://www.mufalani.com.br/blog
        • 3. Re: Too many LISTENER processes (11gR2 Grid Infrastructure (2 nodes))
          Blues Breaker
          Yeah I'm using scan addressing for the cluster and it works.

          I use the 'srvctl status scan_listener' all the time to check it.

          I use this too: srvctl status listener -l LISTENER -n <nodename>

          And it tell me it's enabled and running.

          If I switch my ORACLE_HOME to grid home I can use 'lsnrctl status LISTENER'

          I'm just confused about which lsnrctl to run from where and with which parameter file.

          I got two running but only one has a listener.ora file.

          I guess the one that has a parameter is good candidate but this is contrary to
          what the oracle message is suggesting. It tells me to stop the grid home one and
          run the db home one.

          So should the listener (LISTENER) be running from grid home or db home ?

          Thanks,

          Brian
          • 5. Re: Too many LISTENER processes (11gR2 Grid Infrastructure (2 nodes))
            Mufalani
            Hello Brian,

            In that Metalink note id that I said to check we can see a SCAN DEMO at

            http://www.oracle.com/technology/obe/demos/11gr2/Grid_RAC/11_RAC_SCAN/rac_scan_viewlet_swf.html

            Please seet it to clarify about SCAN feature and check the picture to in this doc about SCAN infrastructure.

            Also check

            http://download.oracle.com/docs/cd/E11882_01/rac.112/e10718/srvctladmin.htm#CDCEECFB

            to manage your listener configuration

            Yes, the (LISTENER) runs at $ORACLE_HOME, and SCAN_LISTERNER inside $GRID_HOME

            Sample:
            The following command adds a listener named listener112 that is listening on port 1341 and runs from the /ora/ora112 home directory on every node in the cluster:

            *$ srvctl add listener -l listener112 -p 1341 -o /ora/ora112*

            Best Regards,

            Rodrigo Mufalani
            http://www.mufalani.com.br/blog
            • 6. Re: Too many LISTENER processes (11gR2 Grid Infrastructure (2 nodes))
              Blues Breaker
              Thanks Rodrigo,

              That was VERY helpful.

              I disabled the old listener and then
              I was able to create a new listener with the correct settings.

              I can see the instances on each node using this
              expression.

              lsnrctl status NEW_LISTENER

              However this shows on the first node that +ASM1, DM3, DM31
              are available on the second node I see only +ASM2 but nothing else.
              I was expecting DM3, DM32 .

              Now when I first created the new listener I read that the 'local_listener'
              setting need NOT be set. When I attemptied to unset that parameter then
              I lost the local instances. I reset the local_listener as before
              and the local instance on node one returned. Not on node two however.

              My question: Do I need to set the 'local_listener' on each instance to the same
              value or to different values to reflect which node they are configured for ?
              I cannot see it now but I remember there is a section that references the
              first node 'node_one-vip' .
              Should the second node's local_listener section be 'node_two-vip' ?

              Now if it need not be set this then something else has to occur to make it work.

              Sorry for not being more brief,

              Brian
              • 7. Re: Too many LISTENER processes (11gR2 Grid Infrastructure (2 nodes))
                Mufalani
                Hello buddy,


                You're Welcome!!! I'm glad that my answer helps u!!!

                Check this documentation about local_listener and remote_listener configuration

                http://download.oracle.com/docs/cd/E11882_01/install.112/e10813/undrstnd.htm#BEIHGBAD

                The Database Agent takes care of this for you --> LOCAL_LISTENER initialization parameters is set automatically.

                Only setup REMOTE_LISTENER to the SCAN port and SCAN, check the doc above for set it.

                Best Regards,

                Rodrigo Mufalani
                htp://www.mufalani.com.br/blog/
                • 8. Re: Too many LISTENER processes (11gR2 Grid Infrastructure (2 nodes))
                  ErikaWhittome
                  Thanks for that link. So what happens when you have a 10g database or 11gr1 database runing on your 11gR2 Grid Infrastructure? Should you have additional listeners named "listener_hostname" on each node running from those 11gr1 or 10g homes?