ietf-smtp
[Top] [All Lists]

Re: retry question

2008-08-06 02:42:15

At 11:38 AM 8/5/2008, Tony Finch wrote:
It says nothing about connections: you can send multiple copies of a
message down the same connection.

I've seen an implementation opening multiple simultaneous connections to do that. It caused resource problems at the server end.

At 03:00 PM 8/5/2008, Hector Santos wrote:
IMV, it depends on the client, is it a MUA client? an backend server list distribution?

MUAs fall under SUBMIT and do not use the same retry strategy as SMTP clients.

In fact, I believe some MUAs will not even continue with the DATA until all same session RCPT are positively resolved. I have to double check if this was because of 5yx responses only, but under ThunderBird if I tried to send to my server which checks for local valid recipients, an address among a few that is invalid, it will not continue sending it until it is fixed or removed from the TO:, CC:, BC: address list. I'll be interested to try this later by forcing a 4yx response instead to see how the MUA will behave.

I expect MUAs to fail if there is any 4yz or 5yz response to a RCPT.

At 03:19 PM 8/5/2008, Hector Santos wrote:
In the end, I think it boils down to how an implementor would design to satisfy the functional specs.

On the automated side:

Does it create 1 copy per recipient on the sender outbound queue?

The implementor also has to take into account recipient limits.

At 03:52 PM 8/5/2008, Hector Santos wrote:
True, but there is at least one MUA - TBIRD that behaves this way. I have not used Outlook in a long time to recall if it had similar behavior. I will lean to the side that it did because its quite possible the latter designed TBIRD researched and borrowed the logic or probably felt it would be easier design for this human usage to make the correction than to try to behave like a real MTA with redundant retries. Although, there is the design option it could issue a prompt:

The difference between MUAs and SMTP client is that the former can revert back to the user for input.

Regards,
-sm
<Prev in Thread] Current Thread [Next in Thread>