-
Notifications
You must be signed in to change notification settings - Fork 1.7k
Description
Please explain the motivation behind the feature request.
I've been running into Bedrock rate limits with Claude Sonnet 4 recently and I'm too cheap to pay for provisioned throughput.
WARN goose::providers::retry: Request failed, retrying (3/3): RateLimitExceeded("Bedrock throttling error: ThrottlingException { message: Some("Too many tokens, please wait before trying again."), meta: ErrorMetadata { code: Some("ThrottlingException"), message: Some("Too many tokens, please wait before trying again."), extras: Some({"aws_request_id": "xxxxxxxxxxxxxxxxxx"}) } }")
Describe the solution you'd like
I'd like to set these values through the config.yaml file.
pub const DEFAULT_MAX_RETRIES: usize = 3; |
Adding an additional bit of delay would help get me past the tokens/min limit.
I have requested quota increase, but it would be nice to control these. 3 tries, 1 sec interval, and a multiplier of 2 seems a little fast for a rate that's measured "per minute".
Describe alternatives you've considered
Limiting the scope of what I'm asking Goose/Bedrock/Claude to do.
Quota increase.
Additional context
- I have verified this does not duplicate an existing feature request
Cheers!
Bryan