PG JGW Process not working Properly, users unable to login to Finesse

[craigers99]

We have Total of 8 PG's in our environment and all are working in Duplex operation, and communicate within the same cluster.

Reported issue: Users Enter their Credentials and Extension on Finesse and receiving "Out of Service Error or below error."
After initial Troubleshooting we have recycle the problem PG (PG1 Service and traffic Shifted to Side B Server and Agent Login immediately happened.) We have done OPC exit on this PG and switch back the traffic to Side A PG. Initially available users logged in and things were working fine. However All other Users attempted to login and received ‘’DEVICE ASSOCIATED WITH THE EXTENSION IS INVALID’’. We have removed the affected users from the PG on the CUCM side and re-added. Initial troubleshooting did not help and we again failed back to side B PG, All users login happened immediately without any issue. So it's only affecting 1 PG currently out of 8.  

Finesse Agent Login Flow:

Agent Login to Finesse , Login ID & Password Authentication with AW Successful, Finesse Submit Login to CTI Server , CTI Server Submit request to OPC , OPC request to PIM for Login request , PIM check Device Target , PIM request to JGW process to add call observer

(Here agent Login is failing and agent reported Device Invalid error after 3 login attempts)

During investigation we have seen below error in JGW logs
handleMsgAddCallObserver::Before addCallObserver for Addr: 11844009 RegState: NotInService.

[bill.king1]

Hi, what version of UCCE and UCM? Also, when you say "We have removed the affected users from the PG on the CUCM side and re-added. " what do you mean you did?

[craigers99]

We are currently running 12.6.2 @bill.king1 sorry to clarify we have removed the device from the Jtapi in programming within CUCM and re-added. 
**Update we are going to attempt a reboot on the Call Manager Subscriber and restart of the PG Services.**

[david.macias]

I feel like I've seen this with UCCE 11.x where the only thing to clear this out was to remove the device from the jtapi user, recycle the PGs (exit opc), then add the device again. After that things just worked.

david

[craigers99 · Accepted solution]

@david.macias  Yep typically how we corrected this in the past but attempted with a few users and nothing worked.  Going to reboot the sub tonight and some PG services, I will post the results after tonight's fun. 

[david.macias]

I would not restart the subs until after you bounce the whole PG. I'm not saying one side then another, but the entire pair at the same time to clear the memory.

david

[craigers99]

Reboot of the PG & Subs corrected our issue.  Thanks everyone for their input!