Once you have a collection of Docker containers running SQLstream s-Server, you can deploy a simple Java application to monitor, through JDBC, how each server is doing. See Installing the Guavus SQLstream Trial Docker Container in the Administration Guide for more details on running s-Server in a Docker container.
This Java application can monitor the pipelines and work as a “heartbeat” for s-Server (a signal indicating that the s-Server in the Docker container is still running) and will be able to tell if an instance is failing or running normally.
In the event of failure, this Java program can create a new Docker container and run all the SQL statements that the one that failed was running and add it to the Kafka partition where the build control node is being made. See Using the Kafka ECDA for Fault Tolerance in the Integrating with Other Systems* for more details on using Kafka for fault tolerance.
For more information on developing an Java application for autoprovisioning s-Server Docker containers, please contact SQLstream technical support.