Question
Highmark Health Solutions INC
US
Last activity: 18 Apr 2016 14:13 EDT
Cache and Rule Resolution Issue on a Multi node environment
We have a multi-node Pega 7 environment referencing a same DB Schema. We have been seeing a weird issue one one node we can see all the changes that we are working on appears intact where as the other node doesnot reflect those changes. We tried to checkin he rules and still face the same issue. Is there any work around or Hotfix for this issue?
Pegasystems
US
There are various things to check, some of which are:
1) Look at SMA on each node to make sure pulse is running every minute
2) Determine what kinds of rules exhibit the behavior. Sections ? (they don't use vtable) Activities ? (they do)
3) Are you waiting a minute, since any quicker and you can't expect the pulse to have run yet
4) What kind of db do you use ? mssql ? oracle ? db2 ? There was an issue fixed that was specific to mssql.
5) Check your db and server time zone and make sure they are the same
6) Something else about a pulse-ms-offset prconfig setting that might be relevant, but I'll let someone else comment more about that
/Eric
Highmark Health Solutions INC
US
We have sent this information to Admin team to work and would update the status soon.
Pegasystems Inc.
US
Hello Dileep,
When you make these changes, is it through a migration or are you checking a rule out in the designer studio? Also, which version of Pega 7 are you using?
Thanks,
Mike
Highmark Health Solutions INC
US
Hi Mike,
I have checked out a rule and updated the rule than checked in as we do for our development based on a use case from designer studio. Am on Pega 7.1.8
Highmark Health Solutions INC
US
Mike Townsend Rajiv Nistala Eric Osman
Now we have identified another issue on top of this happening in a multimode environment.
We recently moved the code to Test Environment and we found that the test environment is not picking the latest rule sets deployed, and its still picking the rules from the Framework.Another issue we also observed is though the system pulse is running and active the last run status is not updated regularly.
Can you please suggest a solution as it is stopping the testing team to go ahead and test their use cases.
We are using the Pega CMF Framework 7.1.3 and the Pega PRPC version 7.1.8.
I would really appreciate if we can get help on this as soon as possible, as this is a high visibility issue that is pending since last 2 sprint development cycles.
Pegasystems Inc.
US
ok, that rules out the specific scenario where the value = 'unknown', which can cause the system pulse issue. Did you check the race-condition scenario mentioned by Eric? see here: https://community.pega.com/support/support-articles/caching-issue-after-deploying-new-rules
Highmark Health Solutions INC
US
I have seen this today and requested the admins to work with this approach. I am wondering if we have any other approach to fix this issues. I will be very glad here if you can explain what is the race-condition scenario and why this occurs and what precautions do we need to take to make sure these scenario wont occur.
Highmark Health Solutions INC
US
Hi Kevin Zheng
We tried setting this prconfig and i still see the same issue. Do we need to do any thing else after we do these setting in prconfig.xml.
we have stopped the server on all nodes
We updated the prconfig.xml
removed the marker file
Started the system.
Still no use. My appliction pics old rulesets though i can see the profile updated.
Pegasystems Inc.
US
what do you mean by 'profile' updated? If this is not production, can you turn on the debug log: com.pega.pegarules.data.internal.access.DatabaseImpl.SystemPulse? If you can consistently reproduce the issue, go ahead with a SR for further investigation, even after all these sanity checks.
Highmark Health Solutions INC
US
Am talking about the Operator Profile by the Word Profile above.
I can enable the debug log for com.pega.pegarules.data.internal.access.DatabaseImpl.SystemPulse and traise the SR with the log file and followup. Hoping this would give any clue for the engineering team to get the Fix for us.
Highmark Health Solutions INC
US
We have stopped one node for now as we don't have enough time to raise an SR and keep the node open and block the developer time. Will keep posted if we enable the node and