34 messages in net.sourceforge.lists.courier-usersRe: [courier-users] Re: MX lookup| From | Sent On | Attachments |
|---|---|---|
| Lukas Vesely | Jul 29, 2002 8:39 am | |
| Sam Varshavchik | Jul 29, 2002 2:20 pm | |
| Lukas Vesely | Jul 30, 2002 7:59 am | |
| Sam Varshavchik | Jul 30, 2002 2:30 pm | |
| Johannes Erdfelt | Jul 30, 2002 3:02 pm | |
| Juha Saarinen | Jul 30, 2002 3:20 pm | |
| Sam Varshavchik | Jul 30, 2002 3:21 pm | |
| Johannes Erdfelt | Jul 30, 2002 3:32 pm | |
| Sam Varshavchik | Jul 30, 2002 5:35 pm | |
| Juha Saarinen | Jul 30, 2002 6:03 pm | |
| Juha Saarinen | Jul 30, 2002 6:13 pm | |
| Johannes Erdfelt | Jul 30, 2002 6:20 pm | |
| Johannes Erdfelt | Jul 30, 2002 6:37 pm | |
| Sam Varshavchik | Jul 30, 2002 6:47 pm | |
| Johannes Erdfelt | Jul 30, 2002 7:20 pm | |
| Tabor J. Wells | Jul 30, 2002 7:31 pm | |
| Sam Varshavchik | Jul 30, 2002 7:46 pm | |
| Juha Saarinen | Jul 30, 2002 8:05 pm | |
| Bill Michell | Jul 31, 2002 1:30 am | |
| Lukas Vesely | Jul 31, 2002 6:51 am | |
| Johannes Erdfelt | Jul 31, 2002 8:43 am | |
| Johannes Erdfelt | Jul 31, 2002 8:48 am | |
| Ben Rosengart | Jul 31, 2002 9:23 am | |
| Moshe Gurvich | Jul 31, 2002 9:32 am | |
| Lukas Vesely | Jul 31, 2002 9:36 am | |
| Lukas Vesely | Jul 31, 2002 9:36 am | |
| Lukas Vesely | Jul 31, 2002 9:36 am | |
| Lukas Vesely | Jul 31, 2002 9:36 am | |
| Lukas Vesely | Jul 31, 2002 10:12 am | |
| Anand Buddhdev | Jul 31, 2002 10:17 am | |
| Johannes Erdfelt | Jul 31, 2002 10:31 am | |
| Sam Varshavchik | Jul 31, 2002 2:41 pm | |
| Lukas Vesely | Aug 1, 2002 10:16 am | |
| Luc Brouard | Aug 6, 2002 12:34 pm |
| Subject: | Re: [courier-users] Re: MX lookup | |
|---|---|---|
| From: | Johannes Erdfelt (joha...@erdfelt.com) | |
| Date: | Jul 31, 2002 8:43:45 am | |
| List: | net.sourceforge.lists.courier-users | |
On Tue, Jul 30, 2002, Sam Varshavchik <mrs...@courier-mta.com> wrote:
Johannes Erdfelt writes:
Although there are other things that will prevent a dead circuit to hotmail.com from plugging up the entire queue, it's still better to minimize the degree of pluggage as much as possible.
First off, what's to say the other MX's at the same priority will be down? You're just guessing here and I don't see how you can guess accurately either way.
It wasn't that long ago when all of microsoft.com was DOSsed off the net because they put all of their DNS servers on the same subnet, and the sole router upstream of their got DOSsed into hell.
A little investigation shows that all of their MXes are in just two subnets. Definitely DOSsable.
Actually, this is the perfect argument for why the behaviour per the RFC's is desirable while the behaviour of Courier is broken.
The way that Courier works guarantees that a transient error reaching one subnet (or DoS, etc) will ALWAYS delay the mail. Whereas just attemping delivery to next same priority MX can get the mail through.
Like I've said, the SAME EXACT benefits for behaviour of NS records applies to MX records.
If all of their servers refuse to accept the mail, then it's their fucking problem.
If you have a huge mail queue as a result of that, affecting all of your mail traffic, it now becomes your problem too.
And you're advocating an algorithm which will guarantee that you'll queue more in the face of network failure?
It's amazing how hypocritical what you're saying is.
So. You have one algorithm where you can queue messages, or you have another algorithm where you can queue messages?
Which is the best one? How about the one that actually follows the standards?
Regardless, even if this wasn't an argument about standards, and this was a problem with causing courier to tie up the queue, shouldn't we fix the problem in Courier rather than wait for this problem to appear in other ways?
The RFC also says you can place a limit on the amount of work you're willing to try.
There's a very big gap between 1 try and 15 tries (for the number of MX's hotmail.com has).
There's a perfectly valid reason for it and no detriments. I don't see how you can just ignore it for anything less than it's horribly broken
You got it. It's horribly broken.
I can't even begin to see how you come to that conclusion.
What you're saying is the equivalent of giving up when you don't get answer from the first root name server.
Sam, seriously, you're being boneheaded here.
Standards are written for a reason, and it's not for you to ignore them because you think you know better without any reason for why the standards are broken.
I've been extremely happy with Courier and your position on this really surprises me.
JE