Re: [ietf-smtp] SMTP Retrying/Sending Strategy on 452 / 4.5.3

Dilyan,

The SMTP-GREY extension fits with RFC5321, Section 4.5.4 "Retry 
Strategies."

    https://tools.ietf.org/html/rfc5321#section-4.5.4

Note the section 4.5.4.1 sending strategy MUST requirement for a delay 
and added considerations for more advanced strategies that can shorten 
the 30 minutes SHOULD:

    The sender MUST delay retrying a particular destination after one
    attempt has failed.  In general, the retry interval SHOULD be at
    least 30 minutes; however, more sophisticated and variable strategies
    will be beneficial when the SMTP client can determine the reason for
    non-delivery.

That is what the SMTP-GREY "retry hint" was intended to do, not get 
into a more complex client/server negotiated retry system where all 
sorts of factors are considered.  The compliant SMTP client MUST delay 
the retry and the extension fits with the extended "sophisticated and 
variable strategies" consideration described above to adjust the retry 
interval.

Beyond this, IMO, it would be a different document like a BCP or 
Information status, perhaps titled:

         "SMTP Outbound Mail Optimization Strategies"

that will serve bulk mail senders in today's environment. It would 
apply in general for remote servers who have deployed greylisting 
methods for load and/or spam controls.

As far as SMTP-GREY, maybe a (draft) clarification for the time-value 
can be appended to section 2.5 "Recommended Blocking Times:"

    .....

    Using a time-value of zero for the reply hint may be used by
    the Greylisting Server in an attempt to convey to the compliant
    SMTP client, the server has no blocking time and the client may
    retry without delay or follow its normal retry schedule.

    Since a non-compliant SMTP client will always follow a normal
    retry schedule, it is recommended the Greylisting Server who wishes
    to reject the first attempt with little to no blocking time for the
    2nd attempt to consider using a retry hint of one second for backward
    compatibility reasons. Using one second should effectively produce 
the
    same almost immediate retry the Greylisting server maybe looking for,
    while using zero seconds may not produce the desired effect of an 
immediate
    client retry and cause an undesired delay, i.e. the client could 
default to
    30 minutes as suggested by RFC531.

Thanks

On 2/16/2019 9:25 AM, Дилян Палаузов wrote:
> Hello Santos,
>
> your draft describes an existing server’s defering strategy and advices clients how to act, when this defering strategy
> is in effect.  One of the essential parts of your draft is writing down the idea (formalizing) for the smtp-clients to
> reuse the IP address on retrying, when grey listing is in effect.  Whether grey listing is in effect, does not depend on
> a 250-GREYLIST or retry= information.
>
> I suggest you include in the draft how a client can recognize that greylisting deferring is in effect and, based on the
> outcome, decide to use or not other IP addresses for subsequent/immediate retries.
>
> If retry=0 is (intentionaly) left undefined, then the draft shall state this, and probably suggest what clients shall do
> on retry=0.  (E.g. ignore).  Keeping retry=0 this way, because anything else is not backward compatible with
> implementations of a draft, does not necessary mean that there is a consensus.
>
> Your draft states:
>     If a SMTP server offers a "retry=time-delay" hint which results in a
>     wasted 2nd attempt and requires additional attempts, the SMTP client
>     MAY begin to ignore the server's "retry=time-delay" hint after the
>     2nd wasted retry.  The SMTP client implementation can decide what
>     limits to place on honoring "retry=time-delay" hints and wasted
>     attempts it provides.
>
> I suggested previously, and this was not tackled, that a retry is not wasted, if the remaining number of recipients has
> decreased during the retry.
>
> Regards
>    Дилян
>
> On Wed, 2019-02-13 at 15:27 -0500, John Levine wrote:
>> In article <12803.1550087591@turing-police.cc.vt.edu> you write:
>>> On Wed, 13 Feb 2019 12:22:09 +0000, Дилян Палаузов said:
>>>
>>>> Is publishing 1024 distinct IPv6 addresses for MX on a domain a good idea
>>>
>>> Only if you're *really* sure that everybody who wants to talk to you supports
>>> at least EDNS0 and doesn't block tcp/53 :)
>>
>> Let me simplify that answer:
>>
>> No, it is not.
>>
>> R's,
>> John
>
> _______________________________________________
> ietf-smtp mailing list
> ietf-smtp@ietf.org
> https://www.ietf.org/mailman/listinfo/ietf-smtp
>

-- 
HLS