2 Replies Latest reply: Aug 23, 2013 10:56 AM by EdStevens RSS

    Grid Infrastructure upgrade introduces invalid LOCAL_LISTENER

    EdStevens

      Oracle DB  11.2.0.2.0 SE-One

      ASM 11.2.0.2.0

      Oracle Linux 5.6 64-bit

      Stand-alone system, no cluster, no RAC

      System has one stand-alone ASM instance and 3 databases registered to Oracle Restart.  One of the databases is not using ASM.

       

      Applying upgrade from 11.2.0.2 to 11.2.0.3, taking the OUI option to ‘Upgrade Oracle Grid Infrastructure or Oracle ASM’, which OUI is selecting as the default.  Everything seems to run fine, no errors, no unusual messages.  But when the process is complete, only the non-ASM instance is restarted.  The two asm-based databases are waiting on ASM to start, and it has failed with an invalid setting for LOCAL_LISTENER.  Examination of the spfile in the new 11.2.0.3 home shows

       

      *.local_listener='LISTENER_+ASM'

       

      However, in the 11.2.0.2 home, (pre-upgrade) the ASM spfile has no LOCAL_LISTENER parameter at all, as confirmed by ‘strings spfile+ASM.ora’.  The running instance (again, pre-upgrade) reports a fully-qualified setting (“LOCAL_LISTENER=(DESCRIPTION=(ADDRESS_LIST=( …. etc)”) but an examination of the alert log for the ASM as well as the db instances shows that is coming from an ALTER SYSTEM SET LOCAL_LISTENER …. SCOPE=MEMORY, as part of the startup sequence. Presumably that is being issued by oracle restart.

       

      I also ran a test of applying GI PSU 11 to the 11.2.0.2 system prior to the 11.2.0.3 upgrade, but it made no difference.

       

      My installations are pretty much right out of the box, nothing fancy.  I can’t find anything about this on MOS.  I posted to the community there and got a reference to two notes that didn’t seem to apply to my situation.

       

      So far the work-around seems pretty simple .. simply remove the LOCAL_LISTENER parm from the spfile and start the ASM instance by hand.  When I did that, the dependent databases also came up with it.  But still, this error seems pretty blatant and I’m surprised I haven’t been able to find anything directly related.

        • 1. Re: Grid Infrastructure upgrade introduces invalid LOCAL_LISTENER
          Srini Chavali-Oracle

          Ed, This seems to be a bug. Can you post a reference to the two notes you mention ?

           

          See these in MOS

           

          Bug 13838047

          737339.1

           

          Have you applied the latest 11.2.0.3 PSU ?

           

          HTH
          Srini

          • 2. Re: Grid Infrastructure upgrade introduces invalid LOCAL_LISTENER
            EdStevens

            Srini_Chavali wrote:

             

            Ed, This seems to be a bug. Can you post a reference to the two notes you mention ?

             

            See these in MOS

             

            Bug 13838047

            737339.1

             

            Have you applied the latest 11.2.0.3 PSU ?

             

            HTH
            Srini

             

            Srini,  thanks for the feedback.  I'd about given up on finding anything on this.

             

            The thread I mentioned that I had started in MOS Communities is in the Storage Management community, titled "invalid LOCAL_LISTENER introduced when upgrading ASM from 11.2.0.2 to 11.2.0.3 "

            The notes I was given there are (Doc ID 1457634.1) and(Doc ID 1543002.1)

             

            Doc ID 1457634.1  speaks of a pre-existing listener and presents a variety of messages found in the various logs. It is not clear (to me, anyway) if this is addressing a new installation or an install/upgrade. In any event, none of the messages reported in the note appear in any of my logs.

             

            Doc ID 1543002.1  talks about a failure to start from srvctl but a successful start from sqlplus. In my case, after completion of the install/upgrade, an attempt to get the ASM instance started from sqlplus also failed.

             

             

            The documents you reference look the closest yet, especially if one pays attention to the caveat "... is a summary description only.  Actual symptoms may vary"

             

            Have not applied the latest 11.2.0.3 PSU, as this is the base, initial installation of 11.2.0.3, using OUI to both install and run the upgrade, so at that point there is no 11.2.0.3 to be patched. Am upgrading only the GI home at this time.  I like to do these things in the smallest increment possible, and so do the GI home and databases separately.   I've never used the OUI option to download software updates as part of the install, but I think I'll try to test that and see what happens.

             

            I am running trials and dress rehearsals on my VBox test lab and have confirmed that the simple work-around is to run the install/upgrade, then manually remove the bogus LOCAL_LISTENER parameter.  I had already tested that before finding the two notes referenced above. While they didn't appear to be a close match to me on causes, the outcome was the same, and they had essentially the same work-around solution .. fix the parameter after the fact.

             

            One thing I did notice in testing was that all of the ASM and db instances in this Oracle Restart config receive and ALTER SYSTEM SET LOCAL_LISTENER .... SCOPE=MEMORY as part of their startup sequence.  Presumably this is being issued by oracle restart, and would override any spfile settings anyway.  But of course if a bad parm in spfile prevents the instance from even starting at all, there is nothing to ALTER SYSTEM against.  I also noted that that ALTER SYSTEM command was coming with a fully qualified spec, not a tnsnames reference.