Created attachment 35947 [details] Patch implementing the desired funcionality This is a patch we're using in production from some years now. This adds a new configuration parameter to load balancer workers, named lb_retries. There can be set the number of retries when a request to upstream fails. Right now the behaviour we observe is that in case of failure the LB retries to all workers at least once. When all backends are busy this adds even more load in our experience. Please consider merging, and ask for any additional info needed. Thanks
*** Bug 48564 has been marked as a duplicate of this bug. ***
Thanks for the patch. I'm testing it now and expect to commit it later today. A couple of minor points so far: - The retries options includes the original request so this new option should have the same behaviour - The patch doesn't include any documentation I've already fixed these locally so this is a suggestion of things to consider for future submissions rather than a request for an updated patch.
Thanks again for the patch. Applied (with minor changes) to 1.2.x for 1.2.44 onwards.
(Sorry for the late reply, just back from vacations) Thanks for the corrections, I'll keep your suggestions in mind for future submitions. Thanks again, F