24.12 Central

In the latest update, we continue improving the resilience and automation of Flussonic Central. Enhancements include running multiple instances in Kubernetes, load distribution, agent reliability, and API optimization. We have addressed key performance and security aspects to ensure your system operates stably even under high loads.

Here are the details of all the changes.

Final Steps for Running Multiple Central Instances in Kubernetes

In this version, we have prepared almost everything needed for the stable deployment of multiple Central instances. As a reminder, running multiple instances ensures system resilience and enables scaling under heavy loads:

• Automatically switches streams to a new instance if a node fails.
• Resolves data duplication issues when transferring streams from one streamer to another.
• Balances the load.
• Provides unified system management and configuration.

We are currently preparing the documentation and will announce when the full capability for running multiple instances becomes available to all.

Automated Stream Distribution to Transcoding Streamers

Streams requiring transcoding are now automatically assigned to servers with this capability.

To balance transcoding, Central considers the stats.transcoder field for streams marked as transcoder and selects the appropriate streamer to handle the stream.

Ensured Agent Connection Reliability

When it comes to security, every second of video matters. In the latest release, we have significantly improved agent stability, minimizing the risk of connection loss. This guarantees seamless archive recording and eliminates situations where a camera is active but fails to transmit a signal.

Documented PGBouncer Integration

When external applications make numerous parallel queries to the database, it’s convenient to use a connection pooler to avoid overloading the database server with separate processes created by Postgres.

To address this, we added a dedicated parameter database_url in the API, enabling the use of PGBouncer to reduce database load.

Accelerated Stream List Retrieval via API

For a large number of streams, the streams_list request could previously take significant time. In the latest release, we optimized the logic for retrieving the stream list, dramatically speeding up the process.

Updated Failover Logic and Improved Health Check

When a streamer fails, its streams are transferred to an active server. Previously, after the streamer came back online, streams had to be manually returned. In version 24.12, we updated the logic: once the streamer is back online, its streams are automatically restored.

Additionally, we have expanded the health check parameters for streamers. They are now also checked for availability, error absence, and correct RProxy configuration. The health check results are available in stats.healthcheck_status.