We currently have an AWS Kinesis Data Analytics app that requires a .jar file to run.
We have automated the deployment for our .jar file that resides in an S3 bucket.
Our issue is, whenever the .jar file is updated we are forced to restart the kinesis app to get the new build which is causing downtime
Does anyone have a workaround or another way of deploying the app Without causing downtime ?
Flink itself does not support zero-downtime deployments. While a few users have built their own solutions for this, it requires implementing application-specific deployment automation and tooling. See
for examples.