Patch Release CX4.4.6

New in CX 4.4.6

Scaling of Controller

Solution deployment architecture has been revamped by allowing multiple replica set of Conversation Controller. Prior to CX4.4.6, only a single Conversation Controller instance could work in a CX deployment that caused slowness in the system, chat stuck issues, and some other unexpected scenarios under load.

With this enhancement, a Controller may now be deployed with a replica set configuration of two or more instances depending upon the expected traffic volume.

QA observed some slowness issues even after adding multiple replicas but the performance is much improved with this scalable architecture.

Clear Tracker in Controller on Conversation End MAJOR

Previously, the Controller cache was not clearing the Conversation Tracker entry from its Redis cache upon Conversation end. This was causing continuous increase of cache size over time.

Now, the Conversation Manager triggers a control event on Conversation End for the Controller to clear the Conversation Tracker from its cache. This ensures an optimal use of the Controller cache memory.

🤬 Fixed Grafana Dashboard Slowness Issue

The Grafana dashboard slows down under load. The dashboard took around 15 seconds to load during a load test with nearly 800 agents and 15 queues. The high volume of data in queue-stats API caused this issue.

In this release, the API is now optimized to return only the required data. The reduced payload of the result improved performance of the /queue-stats API.

CX4.4.6 Release Limitations

Expand To View Limitations

Release Limitations

• Under a load test with 3 chats per second the system experienced slowness after a persistent load for more than 8 hours.

Release Test Highlights

Load test cycles were performed following sanity tests for the planned features.

Load test details. Show