How does the Purge script work?

Technote (troubleshooting)


This document applies only to the following language version(s):

US English

Problem(Abstract)

PV_partition_purge.ksh script will not remove old table spaces, is this
an issue with the script or the environment?

Symptom

Old data files exist even after running the purge script.

Cause


The retention settings and subsequent purge works on data hours, not wall clock hours.

Environment

Solaris

Resolving the problem



Example purge script run:

PV_partition_purge.ksh -dr 40 -da 250 -wa 85 -ma 20

This means,

Delete raw data that is more than 40 days old
Delete daily resource/group aggregates that are more than 250 days old
Delete weekly resource/groups aggregates that are more than 85 weeks old
Delete monthly resource/groups aggregates that are more than 20 months old


The retention settings and subsequent purge works on data hours, not wall clock hours. So if the last data in the system was December 2005, and retention was set to 6 months, say, it'll keep data from June 2006. And this works on a per-channel, per-aggset basis.

So looking at Channel 1 for example:

Table Name Newest Partition Oldest Partition Partition Count Row Count
------------------------------ ---------------- ---------------- --------------- ------------
C01_1DGA_000_H0 2008.10.01-00 2008.09.28-00 4 80324
C01_1DGA_000_H1 2008.09.28-00 2008.03.13-00 200 3780046
C01_1DGA_000_H2 2008.03.12-00 2008.01.25-00 46 700304

C01_1DGA_001_H0 2008.09.30-00 2008.09.27-00 4 244
C01_1DGA_001_H1 2008.09.27-00 2008.03.12-00 200 39472
C01_1DGA_001_H2 2008.03.11-00 2008.01.25-00 45 58439

C01_1DGA_002_H0 2006.06.05-00 2006.06.05-00 1
C01_1DGA_002_H1 2006.06.05-00 2006.06.05-00 1
C01_1DGA_002_H2 2006.06.05-00 2006.06.05-00 1



There are three aggsets, C01_1DGA_000, C01_1DGA_001, and C01_1DGA_002. Take the first set:

C01_1DGA_000_H0 2008.10.01-00 2008.09.28-00 4 80324
C01_1DGA_000_H1 2008.09.28-00 2008.03.13-00 200 3780046
C01_1DGA_000_H2 2008.03.12-00 2008.01.25-00 46 700304

Current data H0 is three days from Sept 28 2008 to October 1 2008, according to the CURRENT_KEEP_DAYS parameter of 3 days.
Recent H1 is March 2008 to September 2008, according to the retention setting of RECENT_KEEP_DAYS of 200.
Archive H2 is about 47-ish days before that, from 2008.01.25 -> 2008.03.12.... add that to RECENT_KEEP_DAYS of 200 makes about 250. Which is the parameter used during the partition purge.

So the system keeps up to the set RETENTION days, based on the last seen data per channel and per aggset. If the last seen data for channel 1 aggset 2 for example is from June 6th 2006, it's going to keep 200 days of daily from that June 6th 2006 date (the data hour data), not from July 2009 (the wall clock hour).

Rate this page:

(0 users)Average rating

Add comments

Document information


More support for:

Netcool/Proviso
DataMart

Software version:

4.4.3

Operating system(s):

AIX, Solaris

Reference #:

1402561

Modified date:

2009-11-13

Translate my page

Machine Translation

Content navigation