After we upgraded application to Pega 8.3.x and 8.4.x from Pega 7 we always asked to do cluster restart in the following sequence due to stream, search services not initialize during the restart otherwise. However because of that we always need to take 2 hrs outage in production. There was infra issue which we need to restart the application daily during the yellow zone. We don’t have a cushion of taking daily 2 hrs outage as the application is franchise critical Pega application for CITI. Pls advise how we can perform a rolling based restart such that it won’t impact the users saying we have all stream, search and web user nodes configured with node based classification.
Restart Sequence we follow today
-Shutdown the application cluster gracefully
-clear the profile temp recursively on the server
-restart search/stream node one by one(we 5 jvms)
-restart batch node one by one (2jvms)
-restart web user nodes in grouping fashion(15 jvms)
***Edited by Moderator Marissa to update Platform Capability tags****