Hey,

You can use options "retries" and "redispatch".
As soon as HAProxy has sent a request to a server, it's not available
anymore in its buffer.
So you're sure it won't be sent again to an other server.

Cheers

On Fri, Jun 15, 2012 at 2:06 PM, Gábor Farkas <[email protected]> wrote:
> hi,
>
> currently i use nginx to handle a configuration, but i'm unable to do
> there what i need,
> so i'm looking into haproxy, if i could use it for load-balancing.
>
> what i'd like to achieve is:
> - i have mutliple backends (let's say 2: backend-a, backend-b)
> - when a request comes in, i want it to be proxied round-robin to one
> of the 2 backends
> - let's say it is sent to backend-a
> - now, if backend-a is unavailable (the port is closed for example),
> or, it's socket-backlog is full,
> i want haproxy to try backend-b
> - haproxy should not try backend-b in any other case. for example, if
> backend-a already started
> to read the request, and then crashed, backend-b should not be tried.
> an error should be sent to the client.
>
> basically i want to be able to failover when the socket-backlog is
> full, but i do not want to double-process
> any http-requests.
>
> is this possible with haproxy?
>
> thanks,
> gabor
>

Reply via email to