How to "push" IUUC_REPL_TABSTG changes to HANA

[justin_molenaur2]

Hello all, I have the following scenario that I can't seem to figure out how to solve.

- Table has been replicated to HANA, all data loads are complete

- I added entries in IUUC_REPL_TABSTG to add two new columns to the source table

- No matter what I do, whether issuing commands from HANA/RS_ORDER table (http://scn.sap.com/community/replication-server/blog/2013/06/24/how-to-start-data-provisioning-witho...) OR using the "update replication status" from IUUC_REPL_CONTENT screen, I cannot get the table structure to update in HANA.

Is there anything I am missing to get columns added to an already replicated table?

For reference, the statement I attempted to use was as follows, with no result.

insert into "DE2_DH2"."RS_ORDER" values('DE2','de2ap01_DE2_10','CE12000',0,'T','COLUMN','N','E');

Thanks,

Justin

[Former Member · Accepted solution]

Hello Justin,

Once you add new fields, you need to click on Calculate new structure there.

Afterwhich you need to drop the data and table in HANA Studio  and start the load/replication again and hence you will see the new structure in HANA.

Regards,

Jomy

[sommerudo · Accepted solution]

Hello,

I agree with Jomy - in all cases where I changed something in IUUC_REPL_TABSTG I had to 'stop replication' and then 'start replication' for the relevant table; the 'start replication' activity then drops the table in HANA and recreates it with the new settings.

Example:

When adding a new field to the table, I had to fill this field for all rows - including the already replicated ones - using a transformation rule; hence I had to load all records again to HANA.

Kind regards, Udo

[justin_molenaur2 · Accepted solution]

Thanks for the responses.

In this case, a table with significant data (billions of records) is already loaded and we would not want to have to drop the data and reload. Is the statement that any structure changes can not take effect until a replication is restarted?

Is there any workaround for this to avoid a reload? Since the key to getting the structure changes into HANA is stop/starting replication, potentially enabling the NODROP feature in IUUC_REPL_TABSTG may satisfy this requirement, any thoughts?

Thanks,

Justin

[tobias_koebler · Accepted solution]

Hi Justin,

there is a way to adjust the table during the replication, but some manuel steps are required.

1.) Stop the master job for the configuration (via LTR or LTRC)

2.) Enter IUUC_REPL_CONTENT and add the new field and a rule if relevant. Save the settings      and execute "Reset Migration Objects".

3.) Enter SE16 and access table IUUC_TABLES.

     - Find your table and open the record

     - Clear the field 'LCL TAB FOR RCV', that signals the system that there is no target table           information and it will be created again.   

4.) Delete the synonym table -> goto LTRC -> Tab "Table Overview" and copy the Logging Table      name

5.) Replace the the first 0 of your logging table with S, e.g. /1CADMC/00001243 will      become /1CADMC/S0001243

6.) Enter SE11 and input the value as database table and delete it.

7.) Enter LTRC and navigate to tab Expert Functions and execute "Reset Replication      Object Definition".

8.) Adjust the table on HANA with the new field.

10.) Restart the master job for the configuration.

This was working for me in my environment. Have fun

Best,

Tobias

[justin_molenaur2 · Accepted solution]

Thanks for the detailed response Tobias. One question here - is step 8 really required? Since we already added the field in IUUC_REPL_CONTENT and you are telling the system to recreate the table, do we really need to manually add the field on the HANA side or will SLT handle this for us?

Additonally, since we don't want to lose any data, should we mark this as "No Drop" in IUUC_REPL_CONTENT in order to preserve data?

Thanks,

Justin

[justin_molenaur2 · Accepted solution]

Worked perfectly Tobias, many thanks.

Regards,

Justin

[timkorba · Accepted solution]

Justin,

In this scenario, wouldn't the added columns automatically come over with SLT and wouldn't you have a massive delta due to the new values in the historical records?  Or are we stating the scenario that you have is only for new records.  Let me know if these steps are still required in the latest DMIS versions as I know this is a conversation from 2013.  Hope all is well.

Tim Korba