Support Article
Pega fail-over does not work in scenario of server crash
SA-10554
Summary
System does not provide continued end-user support when fail-over occurs and user screen was frozen.
During the testing of fail-over functionality found that Pega does not migrate user session to available node within the cluster in case the current node becomes unavailable. Two scenations were tested:
1. The node the user currently is logged in was put in quiesce mode. In this case the user flow continued without problem.
2. The node the user currently is logged in was shutdown. In this case user flow got frosen in one instance and was redirected to Pega login page in second case.
Both tests were executed two times.
The application implementation uses dynamic containers.
Following setting were enabled:
session/ha/crash/recordworkinprogress = true
storage/class/passivation:/rootpath = the directory that is shared across all nodes in the cluster
Error Messages
No error on the screen but user session seems to be hung and need to launch again.
Steps to Reproduce
- Start 2 Pega nodes
- login and identify the node that user session is created
- stop the node where user session resides(no quiescing)
- resume user session by taking some action on the screen
- check whether user session is transferred to other node or not.
Root Cause
The root cause of this problem is a defect in Pegasystems’ code/rules.
Resolution
This issue is resolved by hotfix item HFIX-22194; With HFIX-22194 user session will be rendered in review node after node in case of node crash.
Published June 12, 2015 - Updated October 8, 2020
Have a question? Get answers now.
Visit the Collaboration Center to ask questions, engage in discussions, share ideas, and help others.