Retries & Option redispatch

Previously mentioned retries and option redispatch, with regard to connecting to failed servers.

So "retries 3" and "option redispatch" have nothing to do with the
"inter", "downinter", etc, right? A client connects to haproxy, which
in turn tries to send that request on to a back-end server. If the connection attempt fails (for whatever reason), the docs seem to specify another attempt will be made in 1s (again, to the same back-end server.) If the second try fails as well (again, for whatever reason), the third/final attempt will be sent to a different server (based on the balance algorithm.) In my case, it's balance roundrobin, so the last/third connection attempt is sent to the next server in config file.

The failed connection could be because the backend server is just
"busy" at the moment, it could be because it had failed (but hadn't
yet been flagged as failed), etc.

If the connection succeeds, but say hits the timeout server limit, does that count as a failure and will it retry again?

Thank you,
PH Received on 2010/01/05 20:28