resilience4go

Example

package main

import "time"

func main() {
	bh := bulkhead.New("google.com").WithMaxParallelCalls(40).WithMaxWaitDuration(time.Second * 1)
	rt := retry.New().WithMaxRetries(20)
	newClient := resil.NewClient().WithClient(&myClient).WithBulkHead(bh).WithRetry(rt)
	newClient.Get("http://google.com")
}

This has more human friendly controls. Earlier, we could configure how many threads to use for an external http call. That is difficult to reason about, what does 2 threads mean in terms of cpu, memory?. It is difficult to control, will reducing threads from 2 to 1 reduce the number of parallel calls by half?

Threads are not the right abstraction level when thinking about max number of calls, because they are too low level, the modern OS is a complicated beast. Being able to specify an integer for max number of parallel calls is a high level API and also allows you to change the number by 5% say, which isn't possible in threads.

retry

Bulkhead errors are not retried.