Is your feature request related to a problem? Please describe.
We have noticed that Amazon Bedrock, specifically Claude in AWS GovCloud, sometimes fails with a Service Unavailable error (NOT throttling or quota limit exceeded error). When that happens today, the lambdas do not fail--they hang around until the full 15 time timeout. A solution that just hangs when a service outage or other errors outside of our control occurs does not meet our quality and deployment standards.
Describe the solution you'd like
We would like an internal circuit breaker pattern to be implemented to gracefully handle these failures. We consider this critical for sustainability, cannot deploy without it.
Describe alternatives you've considered
We've considered a lot of things including migrating everything from AWS GovCloud to Microsoft Azure Government (MAG) and using Azure Document Intelligence or Microsoft Intelligent Document Processing instead... ok, just kidding, but we want this feature in a bad way. Pretty please with sugar on top?
Additional context
This determines whether we can release in production sooner rather than later. Not having this feature would have a negative impact to our delivery schedule.
Is your feature request related to a problem? Please describe.
We have noticed that Amazon Bedrock, specifically Claude in AWS GovCloud, sometimes fails with a Service Unavailable error (NOT throttling or quota limit exceeded error). When that happens today, the lambdas do not fail--they hang around until the full 15 time timeout. A solution that just hangs when a service outage or other errors outside of our control occurs does not meet our quality and deployment standards.
Describe the solution you'd like
We would like an internal circuit breaker pattern to be implemented to gracefully handle these failures. We consider this critical for sustainability, cannot deploy without it.
Describe alternatives you've considered
We've considered a lot of things including migrating everything from AWS GovCloud to Microsoft Azure Government (MAG) and using Azure Document Intelligence or Microsoft Intelligent Document Processing instead... ok, just kidding, but we want this feature in a bad way. Pretty please with sugar on top?
Additional context
This determines whether we can release in production sooner rather than later. Not having this feature would have a negative impact to our delivery schedule.