Feature / Question: HTTP Client and API metrics
bsless opened this issue · comments
Is there a protocol or recommendation for metrics for the API client?
Things I'd like to be able to measure:
- Success rate and latency for different ops and endpoints
- HTTP client metrics
- retries
Some of this can be done by wrapping invoke
or providing my own client, but regarding things like retries I'm not sure I can think of a clean way to do it.
Still, seems like it would be better if it was handled by a protocol
Thanks
You can provide a custom retry retriable? and backoff functions without wrapping invoke. See https://cognitect-labs.github.io/aws-api/cognitect.aws.client.api-api.html#cognitect.aws.client.api/client
As for a protocol, it'll be a while before we introduce anything like that given the backlog, priorities, and limited resources, but I'll certainly leave this open for consideration in the long run.