SA-9732

Summary

The user is trying to perform minor skimming after certain number of deployments done.

User reports that skimming process is taking very long time like more than 24 hrs and still shows 6% on the wizard screen.

Even after skimming the ruleset, the ZIP file's size is more or increasing during export of RSV in each iterative deployment. This issue of huge ZIP file size is the issue during deployment like slowness of import/export, environment hang and so on.

Error Messages

Not Applicable.

Steps to Reproduce

Do minor skimming after certain number of deployments done.

Root Cause

This issue of huge ZIP file size is the issue during deployment like slowness of import/export, environment hang and so on.

Resolution

Before starting the skim:
1.Disable agents - simply modify the prconfig entry to be <env name="agent/enable" value="false" />
2.Disable any listeners
3.Disable rule indexing - modify the dynamic system settings to ensure that rule indexing is disabled. The idea being that you don't want unnecessary updates being made even if the indexing agent isn't going to fire.
4.Disable thread dumps - this is more to avoid them from taking up processing time. These do not impact the process.

Things to look for during the run:

If during the first 2 hours you are skimming at a rate less than 90 rules/minute you can go ahead and gather the relevant data and stop the skim. There is no need to run the skim to completion if the skim rate is lower than this. At this point take a look at the logs to check for any issues. If after validating the logs if there is any issues with Application/Web Server or PRPC, you must look to your DBA's for feedback on how the database was performing during the run and if necessary engage the DB2 resources at Pega to assist with the analysis.
The database performance is the primary hurdle that is experienced with skim run.

If the skim is proceeding at a better rate than that for the first 2 hours then you can expect a slow down around record count 14500. This is when the Rule-Obj-Flow records are being processed. Due to the extensive amount of validation as well as the size of the rules the rate falls off dramatically. This continues until approximately record count 18000 when the rate must return to the original value.