OT: GreetPause delay

[Rose, Bobby]

So wouldn't this be the same argument against grey-listing?

-----Original Message-----
From: MailScanner mailing list [mailto:MAILSCANNER at JISCMAIL.AC.UK] On
Behalf Of William Burns
Sent: Thursday, May 26, 2005 12:02 PM
To: MAILSCANNER at JISCMAIL.AC.UK
Subject: Re: OT: GreetPause delay

<sigh>
I can't tell if these responses indicate that some people don't GET this
problem, or that some people don't care about this problem.
I'll try explaining the issue one more time and then I'm gonna shut up.
It's all about the negative impacts of shifting-costs-to-the-sending
MTA.

Let's assume that the "heavy MTA" is running on JISCMAIL.AC.UK.
Let's also assume that the heavy MTA has got more to do than just send
mail to users of the mailscanner mailing list. (it could handle many
mailing lists)

If the subscribers to all these mailing lists have mail admins who turn
on the GreetPause delay, then the performance of JISCMAIL.AC.UK is going
to go to pot.
Now, one particular subscriber of the mailscanner list might not care
that poor list performance causes delays in sending mail on the list,
but the decrease in throughput on the list server may cause an otherwise
altruistic list-serve admin to decide that it's no longer worth their
time to provide free or low-cost list servers to a developer/user
community.

If the typical mail admin starts using a GreetPause, and takes the "they
can contact me" approach, then a list-serve admin will be faced w/ the
prospect of contacting the mail admin of every person who ever
subscribes to one of their lists to make sure that the GreetPause does
not effect machines hosting mailing lists.
This is not going to happen.
Obviously, there are times when paying customers come first, and if
cutting-loose low revenue lists helps, then that will happen instead.

So... Feel free to use the GreetPause feature on your mail server, but
keep in mind that this is not a scalable solution, and you are consuming
more resources on other people's mail servers than is sustainable if
everyone follows suit.

Again, I'd like to mention that greylist solutions provide similar
benefits without this particular liability.

-Bill


John Rudd wrote:

> On May 25, 2005, at 1:54 PM, William Burns wrote:
>
>> For example, a mail server hosting very active mailing list(s) might 
>> easily have to send 10 pieces of  mail per second. If each copy of 
>> the MTA got hung-up for 10 seconds for each piece of mail, then aside

>> from copies of the MTA actually doing work, there'd be another 100 
>> instances of the MTA in memory waiting for prompts.
>>
>> I'm glad that this feature will respect a whitelist in access.db.
>> That leaves open the possibility that someone could at least add on a

>> feature that culls mail logs for good IP addresses, and drops a 
>> whitelist in the access.db file.
>> Without that, it seems like it'd cause a scalability issue for the 
>> mail-carrying internet.
>
>
> Or wait for the heavy MTA that is trying to send you mail to notice 
> "hmm, looks like they're using greet_delay", and they send you an 
> email saying "can you give us an exception?"
>
> I would personally prefer to have such a postmaster _ask_ me for an 
> exception, instead of trying to guess which heavy mail volumes I get 
> are legit and which are not.

------------------------ MailScanner list ------------------------ To
unsubscribe, email jiscmail at jiscmail.ac.uk with the words:
'leave mailscanner' in the body of the email.
Before posting, read the Wiki (http://wiki.mailscanner.info/) and the
archives (http://www.jiscmail.ac.uk/lists/mailscanner.html).

Support MailScanner development - buy the book off the website!

------------------------ MailScanner list ------------------------
To unsubscribe, email jiscmail at jiscmail.ac.uk with the words:
'leave mailscanner' in the body of the email.
Before posting, read the Wiki (http://wiki.mailscanner.info/) and
the archives (http://www.jiscmail.ac.uk/lists/mailscanner.html).

Support MailScanner development - buy the book off the website!