• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Best config for high volume, outbound only mail relay server?

KingGheedora

Diamond Member
Hi,

At work we send out huge marketing email campaigns every day (~1 million / day). We have a middle-tier server that figures out who gets what and forwards the emails through a farm of several linux mail relay servers.

While looking at the logs for the mail relay servers I noticed a lot of messages being sent to the "from" address of the marketing emails. My guess is that these are either delivery delay notifications, and/or non-delivery reports, because the mail relay servers were unable to deliver these messages to the outside world.

What is the best practice in this kind of situation? Should those delivery delay and non-delivery status emails be disabled somehow (and if so, how? these are running Sendmail). The reason i think these are unnecessary is that we parse the SMTP logs from Sendmail to discover bouncebacks in order to clean up our lists, and the from address goes to the same domain as our corporate email system. And the admins from the corp system are getting on our case for the volume of messages being sent to the "FROM" address of the marketing mails. I'm not sure how much of that volume is coming directly from our mail relay farm, and how much is from the outside world (mail that was handed off successfully, but later deemed undeliverable).

Thanks!
 
They may not be SPAM. But your questions are certainly "specialty stuff". Have you asked your external server provider? I doubt that many of us have servers that send out a million messages a day. At least not on purpose. (grin)
 
Not SPAM, we only send to users who subscribe to our service. We run a decent sized website (total 15+ million users). I forgot what the term is for our kind of marketing, where people have actually subscribed for it but probably don't read it. It's some stupid thing that sounds like SPAM, like HAM or something.

Whatever. Our IT dept handles the servers so there is no external service provider. I don't think they have ever handled a situation like this so they won't know the answer.

The size of our email campaigns is irrelevant though. The part I want to know is should those notification emails from our own relay servers be disabled since we are capturing all bounce info straight from the SMTP logs?
 
Is there any reason you need the information? seems like you might want to know if a certain % of your users aren't getting the emails at all or that the IT dept may want to see if theres a common problem on your side that keeps the emails from being sent.

If you dont have a good reason to have that kind of data for anything at all, why not ditch it?
 
Can you write a program that terminates the connection when it sees that the e-mail is being sent to the marketing e-mail and log the From address? I know this can be done using an IIS Event Sink.
 
Can you write a program that terminates the connection when it sees that the e-mail is being sent to the marketing e-mail and log the From address? I know this can be done using an IIS Event Sink.
No idea. I mostly code in Windows environment, and those mail servers where these unwanted messages are originating are linux machines. My guess is that there is probably a simple setting in SendMail where the sending of those status notifications can just be turned off.

Was hoping someone around here might know how this is normally handled.
 
Is there any reason you need the information? seems like you might want to know if a certain % of your users aren't getting the emails at all or that the IT dept may want to see if theres a common problem on your side that keeps the emails from being sent.

If you dont have a good reason to have that kind of data for anything at all, why not ditch it?

That's what i'm thinking. I want to ditch these and I was wondering if anyone here knew if there was any reason not to do so.
 
KingGheedora,

Best practice is to have automated parsers for common delivery status notification emails and to act on them. If you get a DSN for an email address, either remove it immediately or keep a count for removal if it exceeds a threshold.

Surprisingly many sites' mail still accept all email receipients and use DSNs later to bounce errors. Some of them are also large providers who WILL block you if you have too many bounces. So it's in your best interests to act on those bounces and cease sending to them. Not to mention that it's a bunch of network and server load you don't need.

If you're sending 1M messages a day and you are legit, you should outsource it. There's a lot of important best practices here, requiring real coding and network engineering kinds of effort, plus you need to have closed-loop contact with the big email providers to ensure you're following their policies and not getting blocked. Leave it to a pro. That also means your direct mail servers / IP address blocks won't get spam-blocked.

When evaluating an outsourced spammer, start by testing their abuse response. In my experience, many of the big names don't follow-up on abuse reports. They'll make a big song and dance as part of how they justify their legitimacy, but that's just sales and marketing talking, they don't do a thing with actual abuse reports. Oddly enough, many of those same big names' spams are 100% blocked everywhere I have a say in the matter. Fail to resolve abuse reports = blocked. So if you don't want your spam to fall into the same trap, make sure to do a blind test of a potential outsourced spammer's abuse procedures.

Remember, too, any outsourced partner's behavior reflects on your company, so always choose wisely. If done right, you can have them manage a lot of labor and expertise intensive stuff for you so you can get back to doing your real job. If done wrong, you're intertwined with the scum of the earth.
 
KingGheedora,

Best practice is to have automated parsers for common delivery status notification emails and to act on them. If you get a DSN for an email address, either remove it immediately or keep a count for removal if it exceeds a threshold.

Surprisingly many sites' mail still accept all email receipients and use DSNs later to bounce errors. Some of them are also large providers who WILL block you if you have too many bounces. So it's in your best interests to act on those bounces and cease sending to them. Not to mention that it's a bunch of network and server load you don't need.

If you're sending 1M messages a day and you are legit, you should outsource it. There's a lot of important best practices here, requiring real coding and network engineering kinds of effort, plus you need to have closed-loop contact with the big email providers to ensure you're following their policies and not getting blocked. Leave it to a pro. That also means your direct mail servers / IP address blocks won't get spam-blocked.

When evaluating an outsourced spammer, start by testing their abuse response. In my experience, many of the big names don't follow-up on abuse reports. They'll make a big song and dance as part of how they justify their legitimacy, but that's just sales and marketing talking, they don't do a thing with actual abuse reports. Oddly enough, many of those same big names' spams are 100% blocked everywhere I have a say in the matter. Fail to resolve abuse reports = blocked. So if you don't want your spam to fall into the same trap, make sure to do a blind test of a potential outsourced spammer's abuse procedures.

Remember, too, any outsourced partner's behavior reflects on your company, so always choose wisely. If done right, you can have them manage a lot of labor and expertise intensive stuff for you so you can get back to doing your real job. If done wrong, you're intertwined with the scum of the earth.

Tell me more. What are some examples of outsourced spammers? Do you know of any ones that are actually reputable and good at what they do? The only one i've heard of is StrongMail, and they are priced way too high. Something like $60K a month or a year or something ridiculous.

I'm worried about how to actually send the campaigns to my users. We have highly customized campaigning logic that interacts with user data. Keeping 1+ milliion records up to date in some third party campaign system seems like it would be a huge hassle.
 
you have to be bonded to get your high volume mail through to the big isp's. pita to manage.

just run your outbound mail on a couple of qmail servers and mx your inbound mail to a few boxes for that purposes only. keeps the flow going.
 
you have to be bonded to get your high volume mail through to the big isp's. pita to manage.

just run your outbound mail on a couple of qmail servers and mx your inbound mail to a few boxes for that purposes only. keeps the flow going.

What do you mean by "bonded"? I've never heard of this. The only kind of relationship i've seen with the big ISP's is that you can initiate a feedback loop, where you receive copies of all reports from their users who mark you as spam. I signed up for this for AOL and Comcast and we're right now automating a process to unsubscribe the users who initiate these alerts. Yahoo has a feedback loop as well but you need to sign your messages with DKIM or Domain Keys to use it, and we aren't signing our messages w/ that yet. Gmail doesn't have a feedback loop that i'm aware of.

We've been sending to yahoo, goog, hotmail for the most part for over three years. We have issues with yahoo right now though, they periodically greylist us, which is something we'd like to solve. I put a proposal to switch us to PowerMTA or Hurricane MTA since those have much better ability to handle high volume, than Sendmail, and also have domain throttling which might help with the yahoo deferrals. They also have automated bounce processing (those asynch notification type bounces as well as synchronous ones where your messages is rejected during the SMTP session). You just get an output of the bounced recipients from a command line or API and you can clean them out of your user database.

These MTA softwares also support DKIM out of the box so I wouldn't have to code a solution for it myself.
 
I always thought it was a interesting problem. Funny thing is, of course, that there are very few regular "SPAM servers" anymore. Those are blocked pretty quickly by various Blocklists. Apparently the majority of SPAM is now sent by individual malware-contaminated non-server computers.
 
Back
Top