cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
7375
Views
1
Helpful
5
Comments
Arundeep Nagaraj
Cisco Employee
Cisco Employee

FINESSE IN UCCX FAILOVER SCENARIOS

Finesse is the next generation agent desktop available in Contact Center. Rich by its own features, it provides an extensive solution for web based agent desktop.

This short document explains the failover scenarios in Finesse and the expected behavior.

Following are the Service dependencies for Cisco Finesse.

Cisco Unified CCX Cluster View Daemon - Popularly known as the CVD Service, the Finesse Tomcat Service is controlled by the CVD Service.

Condtion: CVD - STOPPED

Impacts:

Finesse Tomcat - STOPPED

Finesse Status Info via the API (http://<ipaddress>:8082/finesse/api/SystemInfo) - NA as there is no finesse tomcat to serve the API calls.

Finesse Administration page (http://<ipaddress>/cfadmin) - NOT ACCESSIBLE as the Finesse Tomcat is down.

Cisco Unified CCX Engine - This is the primary Service for UCCX to cater to calls and agent related activities. This is the core service of CCX.

Condition: Engine - STOPPED

Finesse Tomcat - STARTED

Finesse Status Info via the API (http://<ipaddress>:8082/finesse/api/SystemInfo) - OUT_OF_SERVICE on this node. In case of HA, failover is triggered and the other node will be master and the API will be IN_SERVICE

Finesse Administration page (http://<ipaddress>/cfadmin) - ACCESSIBLE - The cfadmin has Read/Write privileges only from the PUB and Read Only from the SUB as per design.

Cisco Unified CCX Notification Service - Popularly known as the Openfire Service which is responsible for catering the XMPP requests from Finesse, CUIC.

Condition: Openfire - STOPPED

Finesse Tomcat - STARTED

Finesse Status Info via the API (http://<ipaddress>:8082/finesse/api/SystemInfo) - OUT_OF_SERVICE on this node. No failover is triggered.

Finesse Administration page (http://<ipaddress>/cfadmin) - ACCESSIBLE - The cfadmin has Read/Write privileges only from the PUB and Read Only from the SUB as per design.

 

Based on the above dependencies for Finesse, we can conclude the following expected behavior in failover scenarios:

Let us assume a deployment with the following:

Site A = UCCX PUB with co-res Finesse + CUCM PUB + optional CUCM SUBS

Site B = UCCX SUB with co-res Finesse + CUCM SUBS

WAN between Site A and Site B

Finesse Clients are logged into Site A which is the current Master

Failure Scenario 1 - CCX Engine Failover

Site B becomes the master.

Finesse on Site A becomes OUT_OF_SERVICE.

Finesse on Site B becomes IN_SERVICE within few minutes (no specific time duration)

After Site B comes IN SERVICE, the clients are redirected automatically to this node

Recovery - After Site A comes back up and if Site B goes down, Finesse on Site A comes IN SERVICE and the clients are redirected.

Failure Scenario 2 - Finesse Tomcat Crash

Site A remains Master.

Finesse on Site A becomes OUT_OF_SERVICE.

Finesse on Site B remains OUT_OF_SERVICE (by default when Site A is master)

Sessions will be stuck, no operations can be performed.

Recovery - After Finesse Tomcat comes back up, stuck sessions resume automatically.

Failure Scenario 3 - Finesse Tomcat is IN SERVICE but Finesse Status is OUT_OF_SERVICE.

Site A remains Master.

Finesse on Site A becomes OUT_OF_SERVICE.

Finesse on Site B remains OUT_OF_SERVICE (by default when Site A is master)

Sessions will be stuck, no operations can be performed.

Recovery - After Finesse status comes back up, stuck sessions resume automatically.

Failure Scenario 4 - WAN link between Site and Site B is broken (Island Mode/Split Brain Mode)

Both Site A and Site B become Master

Finesse on Site A is IN_SERVICE

Finesse on Site B is also IN_SERVICE

Clients can connect and operate on both the sites.

Recovery - Once WAN link is up, CCX Converges on primary node and becomes the master. Clients connected to Site B are auto redirected to the primary node.

Failure Scenario 5 - CCX Openfire Service crashes

Site A remains the Master.

Finesse on Site A is OUT_OF_SERVICE

Finesse on Site B remains OUT_OF_SERVICE (by default when Site A is master)

Sessions will be CLOSED.

Recovery - After Openfire comes back up, all clients will need to re-login.

 

Comments
ryan_oconnell
Level 3
Level 3

Hello Arundeep,

 

First off thanks for the excellent document.

 

I am presently setting up the exact same topology for a customer using CCX 10.5 with HA over the WAN and Finesse. For some reason Failover never occurs in Finesse. For example if we sign into Finesse with the Agents logged into the Master CCX server, then restart the CCX engine or even the server the agents immediately show that they have lost connection with the master but it never trys to reach out to the secondary server. Only after the original master server is back online and back to master roll will the agents connect to the original server. Is there something special that needs to be configured in the Finesse server that lets the clients know that there is a second server to connect to?

Arundeep Nagaraj
Cisco Employee
Cisco Employee

Hi Ryan

This is a known defect on 10.5 versions where the secondary node is missing a passphrase file during the install which will prevent the failover for Finesse.


https://tools.cisco.com/bugsearch/bug/CSCur00799/?reffering_site=dumpcr


The workaround needs to be done using root, so you will need to open a TAC case with us:)

Regards,

Arundeep

ryan_oconnell
Level 3
Level 3

Thank you I'll start a cisco Case 

ryan_oconnell
Level 3
Level 3

Hello Arundeep,

 

A few months ago you helped me with this issue, and you pointed in in the direction of a defect which indeed was the case and resolved my original issue.

We do have a new issue that I thought you may be able to help with as well.

Now that failover is functional, my customer who runs CCX 10.5 with Finesse was doing some failover testing. It seems as though clean failovers work just as they should, as you have described above. My customer added one more test. They wanted to simulate a server crash. This is what they did.

To simulate a crash I disconnect network card from uccx server.  I was logged into finesse at the time but did not received the banner saying failover has occurred.  I check CCX in the alternate data center and it was showing Master but I was not able to do anything from finesse.  This is the first time I notice this behaviour, other time it shows from that failover has occurred and that I connect to CCX server at the alternate datacenter.  

I read through your scenarios above trying to understand if manual intervention is required for this type of failover to occur. Since it wasn't a clean failover does the customer need to do something in order for the finesse agents to failover?

 

 

 

 

 

Hello, 

 

thank you very much for that extension. 

 

I have just one question with another scenario that I am looking for an answer. 

 

Uccx ver 11 HA 

 

SITE A -ACTIVE 

finesse agents connecting to this server. when i disconnect the network leg from that server (a scenario that the switch connect the VM that the uccx in it in down) the agents are disconnecting. the server is up but with no network. so the uccx sub not doing nothing. I that it is island mode . can you share your Thoughts on that mater? 

 

thanks a lot .sharon carasenti.

 

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: