Rolling restarts are not transparent to active user
SummaryAfter High Availability is configured, existing users logged in are not transparently passed to the active server during a rolling restart scenario.
System Management Application (SMA) shows that the quiesce processes completes successfully after a server is disabled via the load balancer, and no active requestors are shown on the server being brought down.
Users experience unending load screens when taking actions and receive the error "PegaRULES Request Status error Too many identical requests in quiesce mode." when user attempts to close an open work object.
Error MessagesToo many identical requests in quiesce mode. Please see your system administrator. Unauthenticated or not available No ID available Unknown.
Steps to Reproduce
- Configure shared storage and update passivation path in prconfig.xml file.
- Apply redundancy across the infrastructure.
- Configure Load Balancer with Cookie persistence and give cookie name as 'cookie'.
- Make changes in prconfig file.xml per High Aavailability guide.
- Log into one of the server in cluster as a user (Non Quiesce admin).
- Disable the corresponding server in load balancer.
- Begin Quiesce on the server.
- Try to open work objects through the portal.
Root CauseThe root cause of the issue is due to a configuration problem in user's operating environment.
The cookie name was configured incorrectly.
Also, there was a one to one mapping between the Webserver and Application server in the environment used for testing high availability.
ResolutionThe following options are recommended:
- Update the cookie name to JSESSIONID.
- In-order for the routing to work, at-least two nodes should be mapped to a webserver. One can have multiple web servers, but each one needs to have more than one node to be routed to (as defined in the plugin-cfg.xml).
Published October 5, 2016 - Updated October 18, 2016