On Feb 25, 2010, at 14:54, LuKreme wrote:
On 25-Feb-10 00:22, Brian Willoughby wrote:
Their site times out after one minute of waiting for a response to an
SMTP connection, and considers the destination to be invalid at
that point.
a 60s timeout on SMTP is too short. The recommend is... erm, 15
minutes comes to mind though I would argue that is too long anymore.
5m is certainly defensible. I've seen servers with timeouts set to
15s, but those are run by people who really don't want to receive
email from the world at large.
This is very helpful. I had no perspective. I get impatient waiting
60 seconds for manual SMTP connections via telnet when testing, so I
couldn't immediately shoot down their limit. I did speak to their IT
rep once, so I might call back and suggest that they increase the
timeout a bit.
It's only recently that I called a client on the phone to find out
why they had not responded to my email, and that's when I learned
it had been bounced by their system.
Erm, BOUNCED implies they RECEIVED the mail, ACCEPTED the mail, and
then decided they couldn't deliver it, so bounced it back. In the
scenario you described, the message should have been REJECTED. If
they did, in fact, BOUNCE the message that is a bad sign.
Two things: I'm probably being sloppy with terminology, and I perhaps
didn't explain the one situation where there's a problem.
I sent mail to a business, they received my mail, they replied to my
mail, but I never received their reply. When I called, they claimed
to have sent the email. I'm not sure whether they got an error
message back from their mail server, or if they only found out the
"why" after specifically asking their IT for more info.
If the message was REJECTED, the question would then be how could
you not know? Don't you check your logs for problems like this? Or
is this server basically personal? If you're handling mail for
anyone other than yourself, you should have a log analyzer that send
you mails showing any issue that might need attention. I run
postfix, so my log analyzer is pflogsum, which I run via cron every
time the log rolls.
Assuming we're talking about the same thing, I'm not sure whether I
would get an error message. If they connect to my sendmail via SMTP,
but then give up before even trying to send mail, I don't know what
sort of error to expect. I have seen error messages showing spam
senders who connect and leave the connection open without ever sending
mail or using QUIT.
To answer your question, this is basically a personal mail server. I
do have a few friends who have free accounts. Basically, I wait for
complaints to grab my attention, and then look in the log files for
clues to fix whatever issue there might be. You have a good point
that I should run an analyzer. I've turned on various logging
options, turned off others, and possibly even changed which log files
store certain levels of errors. But I suppose any analysis tool
should be prepared to deal with the options that are available. I'd
ask for recommendations, but you already say what you're using for
postfix, so I should probably just search for something in the
sendmail family.
Bottom line: Based on your response, I get that impression that so
long as my server responds within 5 minutes, I should have a valid
case to complain that they're blocking their own company staff from
contacting me (their customer) and potentially many others. I'll
probably call them and suggest 15 minutes so I can negotiate down to 5.
Meanwhile, if anyone thinks that I should be able to expect a response
in under 60 seconds - and has suggestions as to how I might rectify
this - then please follow up with me. But I'm leaning towards
considering this to not be a problem on my end.
Brian Willoughby
_______________________________________________
MacOSX-admin mailing list
[email protected]
http://www.omnigroup.com/mailman/listinfo/macosx-admin
