Disk space when running RSXM_TABLE_SWITCH (Switch deletion procedure)

[Former Member]

Hi Forum,

We have more than 4.000.000 processed messages in XI we want to delete. Our interfaces are Archiving active and Database is running in Oracle 10.2.0.4.0.

We've mark the SXMB_ADMIN->Configure delete procedure->Switch Procedure Activate checkbox. Now, container 2 is the active container.

Then, we started the copy job running program RSXM_TABLE_SWITCH. We did not notice the tablespace PSAPSR3 had only 20GB of free disk space when we run the job. So, copy 4.000.000 messages to container 2 need much more than 20GB. We've cancelled the job.

Since we canceled the job, we can't run any other PI deletion job until the restart of the canceled job. We've made some search in this forum and seted parameter DELETION->DROP_MAX_TABLE_LOAD = 10% and parameter DELETION->TABLE_SWITCH->MOVE_INSTEAD_OF_COPY = 'X'.

Now, we restarted the job, and it is still consuming the disk space. Container 2 tables are growing and container 1 tables are not releasing disk space.

What can we do to delete the messages? is any way we can made a rollback the table_switch set container 1 active again an delete the messages before the switch? Any Oracle command, like Alter table?

Thanks in advance,

John.

[Eoin_Kierans · Accepted solution]

Hello John,

When using the Switch procedure only messages which are inside the retention period or are not in a final status will be copied across. All other messages should be marked as logically deleted, i.e with a REORG value of DEL within the table SXMSPMAST, how many of these there are can be checked in SE16.

Therefore if you wish to delete these messages which will not be copied across in advance of the switch procedure completing you can run the report RSXMB_SWITCH_DEL_OLD_ENTRIES. This report allows you to physically delete the logically deleted messages, this can be done while the switch has been cancelled, there is no need to roll back the switch to run this report. I would then recommend carrying out an online reorg of the PI tables affected by the switch to release the disc space to the OS. You can see the list of these tables within the following wiki:

Overview of the Switch Deletion Procedure

You can also check the following KBA's for further info:

1788321 PI: Switch job is canceled due to lack of disc space

1705321 PI Deletion job fails with error 'Repeat terminated copy job first'

1744613 Archived messages are not physically deleted from the PI tables

1749780 Changing the Maximum Number of Table Entries in the PI Master Tables

Regards

Eoin