Preparation migration with Unicode

[Former Member]

Hello,

I need to migrate the BW system to Hana, and before that I need to execute preparations steps for unicode migration.

I'm checking for Cluster Tables but I have some doubts about how to manage the results from SDBI_CLUSTER_CHECK report.

I've only records on tables CDCLS, CVER4, DOKCLU and EDI40.

 

This is the explanation from the Configuration Guide

Check and analyze inconsistencies in Cluster Tables. Report SDBI_CLUSTER_CHECK can be used to search for initial cluster records in cluster tables. Initial cluster records must be removed before the database export, otherwise the export might cancel. SAP Note 1348055

  Initial cluster records for Unicode migration provides a description of the report and explains how to proceed when initial cluster records are found. More information about the latest version of the report can be found in SAP Note 1077403 Cluster table check using SDBI_CLUSTER_CHECK

Initial Cluster recors are the entries from cluster records checked? If yes, in which phase from DMO I need to delete this entries?

Thanks a lot.

[Former Member]

Hi Ruben,

There is no error reported in the cluster check you ran which is good. As the screen shot mentions you are good to proceed with subsequent steps.

Few informative links regarding your current migration, which you may have already checked.

1309804 - SDBI_CLUSTER_CHECK: Operation and error messages

https://scn.sap.com/thread/1867589

Thanks,

Manas

[Former Member]

Thanks Manas,

Then in this case is not necessary to delete the records because there aren't incorrect entries.

Initial cluster records means incorrect entries on this cluster tables?

Regards

[Former Member]

The answer to your question is in Note 1309804 - SDBI_CLUSTER_CHECK: Operation and error messages

• ERR_INI - An initial cluster record was found.

                    An initial cluster record is a database line with the values PAGENO = 0 and PAGELG = 10.
                    Temporarily, initial cluster records may occur for databases that work in the dirty read mode; they are written by transactions that create new physical cluster records to be able to hold the relevant locks on the database. However, these initial cluster records must have been removed from the database at the latest when the database transaction is exited using COMMIT or ROLLBACK. An initial cluster record whose generating transaction has already been exited using COMMIT or ROLLBACK must always be classified as an incorrect cluster record as a result. The check report reports an error if the initial cluster record exists longer than 15 minutes.
                    For databases that work in the consistent read mode (Oracle), the temporary initial cluster records remain hidden to other transactions. In these cases, initial cluster records have to be classified as errors. Therefore, the system reports all initial records that are found as being incorrect.
                    A large number of initial cluster record is related to problems that have been solved by Notes 456925, 456718, 456682, 456682, 456576, 456561, 456553, 456546, 455769, 424339.

There can be other errors as well which are reported in the same Note.

Thanks,

Manas

[Former Member]

Sorry for insisting

But the number of records checked is the same that Initial cluster records or these are only the records that are incorrect ?

Thanks for advance

[Former Member]

No Problem

No of record checked is the total number of records checked from that Cluster table. In case of any error (initial or any other) it will be reported by SDBI_CLUSTER_CHECK. Different errors have different codes (which is described in the SAP note) and eventually needs to be handled accordingly.

I just ran one for EDI40 in my test system and below is the result. The initial records and records with error are shown separately.

Thanks,

Manas

[Former Member]

Hi Ruben,

Were you able to go past your issue?

Thanks,

Manas