I had opened an earlier post on this, but closed it thinking that our db team had provided a working solution. The solution is problematic (time wise) for the production environment, where the table has 98 million rows.

Here is the scenario:

In 6.3, the pxattachkey column is varchar2. We are making use of multiple frameworks including CS, CSFS, FSIF, SCM. PegaCall

In the framework installation process for CS, one of the sql statements generated is

ALTER TABLE SSWMD01.pc_data_workattach MODIFY “PXATTACHKEY” DATE This fails because there are values in the colum.

The database is Oracle.

I have two questions:

1. what is the impact of removing this statement and just leaving the column as is (varchar2)?

2. If there is an impact, what is the most efficient method of modifying a column in a table with 98 million rows?

Our DB team kicked off a test against a copy of the prod database and eventually killed the process (after 5 hours of running). The were using

"create table as select * from table" as the method

Is column order important for the pc_data_workattach table? Our DB team has indicated that there is a relatively fast solution if the column order within the table does not matter. Basically, they would create a column at the end of the table and update/convert from the pxattachkey column. They would then drop the pxattachkey and rename the new column (or something along those lines). Anyway, it could only be done if column order is not important to Pega.

***Edited by moderator, Maryrita: moved to Applications from Upgrade Center***