When will a sending mail server fallback to the receiving domains backup mx?
Hi all,
I have a mail server that operates a couple of gateways as a backup mx for a couple of clients (necessary due to their ISPs being very flaky), recently their Exchange server became corrupted (so the internet connectivity was still in place). During this time no messages were delivered to the backup gateway.
I've seen many circumstances where - if a mail server is playing up and not simply unavailable, sending servers will not attempt delivery to the backup mail server - does anyone have a definitive list of the circumstances? For example:
When sending server gets EHLO response but no delivery, backup mail server not contacted
Emails go to the lower number mx records you have listed. So if exchange is 5 and then your backup is 10 when the exchange server is not available it should forward to your backup. Maybe your exchange server is just eating those emails? Turn it off and see what happens.
Thanks for the response - believe me I understand DNS inside out (not by choice!).
My question really is exactly what defines "when the exchange server is not available".
For example, if:
A sending server attempted connection to IP: OK
Through firewall: OK
Says EHLO to mail server: OK
Attempts Deliver: FAIL
Will the sending server retry to the primary (given that it is online and responding) or will it fall back to the backup MX.
I'm 99% certain it will retry to the primary until if gives up because the server is there, it's just being problematic. As I understand MX doesn't operate at the *message* level, it operates at the *connection* level.
Alas I can't do further testing as this is all stuff that has happened in the past to a clients Exchange. Our backup gateway didn't have any messages for them and they are stamping their feet about it yet we can see all our systems are working fine.
Generally, the sending server tries to deliver mail to the first mx from the returned list that is responding
meaning that if primary mx record is not resolvable or unreachable (unreachable at all or non-responsive on port 25) then the backup MX is used
If primary MX is "responsive" (playing up) - i.e. sending server can connect on port 25, but then gets no answer/timeout - then the backup is not used.
Same goes if primary responsive and reject mails etc - the backup is not going to be used
the only exception (that I know) to the "responsive" rule is made to certain 4xx SMTP return codes. Although it may not be implemented on all mail servers, it could be that when sending server connects to primary mx and receives smtp response 4.2.1 or 4.5.1 or 4.5.2 it would go to the backup mx record
>Alas I can't do further testing as this is all stuff that has happened in the past to a clients Exchange. Our backup gateway didn't have any messages for them and they are stamping their feet about it yet we can see all our systems are working fine.
Do you not have the historic transcript logs you could trawl through?