DNS servers already return multihomed host IPs randomly.
So I don't think that a similar algorythm would be so useful.

Multihoming is not used for clustering, but for loadbalancing.

In real use cases you won't get any real advantage by remembering the failing multihomed ip in a following retry (being done minutes or hours later).

Furthermore, by specification multihomed IPs MUST NOT be tried in a "deterministic way" but "randomly".

As an example: if you have a multihomed HTTP host, your browser does a single request to a random IP of that host and if it fails the browser return the error. Multihomed ips are not checked upon failure. I don't understand why SMTP would need a different behaviour than HTTP, considering that SMTP already provide failover mechanism via MX support while HTTP doesn't even have a "protocol defined" failover mechanism.

Stefano

Martijn Brinkers (List) wrote:
You may be interested in a patch I wrote:
http://issues.apache.org/jira/browse/JAMES-358

Unfortunately the patch has not been accepted.
I still think that it should be applied at least as an option.

The problem is that each mx record and all accompanying hosts should be
tried at least once and preferably in a deterministic way. Looking at your
patch it seems that only one ip is tried. What I think would be better is to
'save' (a property of the mail object) the last tried ip address. If a
connection to the host fails (for whatever reason) the mail should be placed
back in the store and the next mail should be tried. Later on when the
failed mail is retried it already knows which ip address it has tried and
skips to the next ip address to try. Using an approach like this does not
give a host with many ip address an uneven 'advantage' over other hosts and
makes the system more responsive if some domain somehow blocks incoming
email (Hotmail has a large number of hosts).
Cheers

Martijn Brinkers


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to