I am trying to deploy Rasa on AWS Fargate. There are two tasks running on Cluster. One is Rasa-Server and another one is Rasa-Action-Server.
The problem is when increase the load Rasa-Server and Rasa-Action-Server keep restarting. I have configured 4 vcpus and 16GB for memory.
Is there any recommend guideline to explain how we can deploy rasa on Fargate. Or is there any configuration changes for Rasa when production deployment.
I’m wondering if Fargate assumes a
/alive endpoint for it’s containers that Rasa doesn’t offer. I’m not familiar with Fargate but most cloud tools require this. @Tharanga, can you confirm what requirements Fargate containers must have?
Thank you @koaning . I configured the action-server /health endpoint for AWS ALB health check. Restarting issue was resolved after that.
Then I got another issue. I am using AWS DynamoDB as TrackerStore. I got exception
ProvisionedThroughputExceededException. This is cause of Read/write capacity limitation from the DynamoDB side.
I changed that capacity mode into
On-demand mode. Now it’s working as expected.