8 Replies Latest reply: Jun 4, 2012 10:20 AM by 849425 RSS

    advice on final step of migration: steps to stop ggs and cutover to target

    849425
      I have tested migration couple days ago. So far the extract/replicat still running.

      What is the proper way to cut it over to target?

      Should I stop all ggs processes on source first, then stop replicat?

      On target side, I have configured 20 replicat processes to match 20 extract on source.

      How do I be sure all the replicat finished on the same timestamp? same scn?

      Is there any offical oracle documents I can check into it?

      Thank you for all your response.
        • 1. Re: advice on final step of migration: steps to stop ggs and cutover to target
          amardeep.sidhu
          Should I stop all ggs processes on source first, then stop replicat?
          Yes but you need to make sure that no DMLs happen on the source side once you have stopped the Extract.
          On target side, I have configured 20 replicat processes to match 20 extract on source.

          How do I be sure all the replicat finished on the same timestamp? same scn?
          Not a problem. When you have stopped the Extracts on source side, all the replicats are going to process whatever data they have received and then just run idle.

          Basically you need to stop Extracts first then wait for the Extract Pumps (assuming you have separate Extract Pump processes to transfer trails) to transfer everything to the target side. Finally when Pumps are done with transferring & Replicats with applying those changes you can point the things to the new database.
          • 2. Re: advice on final step of migration: steps to stop ggs and cutover to target
            849425
            Thank you ver much. this almost answered all my question.

            Last question here is: are all the replicat process will apply changes to a one point---like same rbA or same scn?

            any views i can query to finalize the cutover time?

            thanks.
            • 3. Re: advice on final step of migration: steps to stop ggs and cutover to target
              amardeep.sidhu
              846422 wrote:
              Thank you ver much. this almost answered all my question.

              Last question here is: are all the replicat process will apply changes to a one point---like same rbA or same scn?

              any views i can query to finalize the cutover time?

              thanks.
              That depends upon how exactly you have configured multiple Extracts ? Are you using RANGE function or just divided the tables into different Extract groups ?
              • 4. Re: advice on final step of migration: steps to stop ggs and cutover to target
                849425
                I just divided tables in different extract group.
                • 5. Re: advice on final step of migration: steps to stop ggs and cutover to target
                  amardeep.sidhu
                  It is pretty simple actually.

                  Stop all the DML in the source. Wait for sometime so that Extract's capture the pending changes, if any. Check the lag at Extract end and stop them.

                  Now do the same with Extract Pump

                  & finally Replicat.
                  • 6. Re: advice on final step of migration: steps to stop ggs and cutover to target
                    849425
                    ok, great. It seems those are the only steps I have to run. Thanks. I will test the results today , if all works, I will close this thread.

                    Thanks again.
                    • 7. Re: advice on final step of migration: steps to stop ggs and cutover to target
                      849425
                      Hi, Armardeep:

                      Can you take a look at those datapump extract, are they mean no laps?

                      GGSCI (oracle5) 37> info pother

                      EXTRACT POTHER Last Started 2012-06-04 15:05 Status RUNNING
                      Description Data Pump, reads from ./dirdat/eo
                      Checkpoint Lag 00:00:00 (updated 00:00:01 ago)
                      Log Read Checkpoint File ./dirdat/eo000005
                      2012-06-02 03:01:11.831680 RBA 1019


                      GGSCI (oracle5) 38> info plob

                      EXTRACT PLOB Last Started 2012-06-04 15:05 Status RUNNING
                      Description Data Pump, reads from ./dirdat/eb
                      Checkpoint Lag 00:00:00 (updated 00:00:05 ago)
                      Log Read Checkpoint File ./dirdat/eb000005
                      2012-06-02 03:00:59.896792 RBA 1015


                      GGSCI (oracle5) 39> info pclob

                      EXTRACT PCLOB Last Started 2012-06-04 15:06 Status RUNNING
                      Description Data Pump, reads from ./dirdat/ec
                      Checkpoint Lag 00:00:00 (updated 00:00:09 ago)
                      Log Read Checkpoint File ./dirdat/ec000005
                      First Record RBA 2353


                      Here are final moment of replciat:

                      GGSCI (oracle07) 2> info rclob

                      REPLICAT RCLOB Last Started 2012-06-02 02:57 Status RUNNING
                      Description "Change data apply, reads from ./dirdat/pc"
                      Checkpoint Lag 00:00:00 (updated 00:00:09 ago)
                      Log Read Checkpoint File ./dirdat/pc000002
                      2012-06-04 15:06:10.109634 RBA 1150


                      GGSCI (oracle07) 3> info rlob

                      REPLICAT RLOB Last Started 2012-06-02 02:57 Status RUNNING
                      Description "Change data apply, reads from ./dirdat/pb"
                      Checkpoint Lag 00:00:00 (updated 00:00:09 ago)
                      Log Read Checkpoint File ./dirdat/pb000003
                      2012-06-02 03:00:59.825526 RBA 1250


                      GGSCI (oracle07) 6> info rother

                      REPLICAT ROTHER Last Started 2012-06-02 02:57 Status RUNNING
                      Description "Change data apply, reads from ./dirdat/po"
                      Checkpoint Lag 00:00:00 (updated 00:00:10 ago)
                      Log Read Checkpoint File ./dirdat/po000002
                      2012-06-02 03:01:11.743623 RBA 1213


                      I wonder on replciat side, why let's say the replciat for clob tables, why it is reading from File ./dirdat/pc000002 but pump information shows it is : File ./dirdat/ec000005?

                      Is that means I have to run the replcat a while until to the 00005 pump file?

                      Thanks
                      • 8. Re: advice on final step of migration: steps to stop ggs and cutover to target
                        849425
                        I have stopped extract process on source, but left datapump processes running.

                        So is it I have to wait for all the last ./dirdat/00000seq# to be showing in replcat information before shutdown datapump and then let replcat apply changes, then stop the replicat?

                        Thanks