When run dual mode , agents forced logged out ...

[dolphine55]

I wonder if some one had this experience with UCCE dual mode implementation.

We have a system which has implementated with dual mode UCCE.  There are 2 PG site A and B.   Supposedly both PG should be running paral for fail over purpose.   However for some reason , when 2nd site turned on.    All agent will be forced to log out, which cause service down.

I'm not very experienced with this implementation, but believe that some kind of confliction might have happened on the implementation.

I wonder does  anyone have this experience ?   Where should I start to trace this issue ?   Any direction will be greatly appreciated.

Tina

[geoff]

I don't know why you call it a "dual mode implementation".

It sounds like the PG and/or CTIOS is not set up correctly.

Bring up the PG on the A side and its JTAPI gateway (assuming you are using Call Manager) will go active, as will the PIM. Start the CTI server on A and it will be active. Bring up the B side PG and CTI server and they will be idle.

Bring up the CTIOS on the A side. You will see trace in CTIOS that it's looking for its peer on the B side to connect to. This trace will keep repeating as it tries to talk to the B side CTIOS. Bring up the B side CTIOS and the trace will stop and both CTIOS will be active - see title bar.

If this does not happen, ensure that the NIC binding order is correct - public first, private second. If this is wrong, the peers won't communicate and you won't see them stabilize and both show active in the title bar.

Now log in an agent. The CTIOS client will be connected to both CTIOS, choosing one as a primary and the other as a secondary. The client chooses at random which should be primary. Look at the client - it will show which one it is connected to. Stop that CTIOS - the client will show it switching to the secondary. This must be seamless and not log you out.

If the CTIOS is working correctly and bouncing CTIOS does not disturb agent logins, move onto the PGs. Cycle the PG on the A side - the B side should immediately become active. Agents will not be affected.

If this is not what you see, check that the setup of the PG is correct - that the private connections are correct and that you can ping the other private side from each. Do a tracert to make sure the private traffic is not going through the public interface. If it is you need to add persistent static routes to address the single default gateway problem in Windows.

Regards,

Geoff

[dolphine55]

Hi Geoff,

Thanks for your prompr reply and very detail explanation of the trouble shooting steps.   It is very helpful , and I wonder if I have found the answer of my trouble  here already :

From your message I Iearnt that CTIOS is running side by side to offer service to the Agents.    However in our system environment,  I find that 2 CTIOS has been implemented with the same name, ie.  rather with name_A, name_B, both servers on site A and B are using exactly same name for instance: CTIOS1.     I think this is the culprite that causing all agents knock out.   As agents are looking for idle the connection for one server but either one they choose will causing both connection down as they are using the same name.    Am I right Geoff ?

Now my second question is : I was told that the CTIOS name was given upon implementation and can not be changed.    However if you agree with my assumption that the name is causing the issue.   What is the step to change the CTIOS Service name?  

Thanks so much and I really appreciate your insides !

Tina

[geoff]

However in our system environment,  I find that 2 CTIOS has been implemented with the same name, ie.  rather with name_A, name_B, both servers on site A and B are using exactly same name for instance: CTIOS1.   

I am reluctant to say anything without seeing a screen shot. You don't quite use the same words I do (e.g. "

dual mode implementation") so I am a bit wary.

With a duplex install of the PG, PGA and PGB are very similar. CTIOS would be running on both and it would be called CTIOS1. I don't quite know what you mean by "2 CTIOS has been implemented with the same name, ie.  rather with name_A, name_B, both servers on site A and B are using exactly same name for instance: CTIOS1. ".

I would not expect to see two CTIOS servers on PGA though.

The client connects to PGA on port 42028 and PGB on port 42028 by default. Do you use these values?

Post a screen shot of ICM setup on PGA.

Regards,

Geoff