mx record priorties.. correct? exchange failure - no bounce backs?

[markm75]

We have two internet providers at our location where our exchange server resides...

call them wan1, wan2

We also subscribe to the free backup mx service of rollernet and their two domain names.. call them rollernet1.us and rollernet2.us...

In our DNS settings we have wan1 and wan2 as the first two mx records both with the same priority of 10

The two rollernet addresses are the next two inline with values of 15 and 20.

Does this configuration sound right?

IE: if our domain/location goes offline (wan1, wan2).. rollernet would pick up the mail and queue them or would either of our isp's first hold the mail then release them and get picked up by rollernet?

We recently had a complete hardware failure of our exchange system..

At that point i actually had wan1 set to 10 and wan2 set to 15, then the other two after that..

For some reason, during the period we were done (and i was out of town).. over say 18 hours or more.. noone has received bounce messages on the outside.. nor has anyone inside the building received those messages.

The only thing i was able to note, is that once i had systems back up.. rollernet sent us a message like this:
------------------
[rollernet] Mail domain 'youromain.com' disabled

An SMTP rejection was detected while attempting to deliver mail to 'yourdomain.com'. In accordance with our backscatter prevention policy, our mail servers will stop accepting messages for this domain. To view errors related to this problem, view the "SMTP Bounce Logs" under Mail Logs in the account control center. Once the problem has been corrected, you may reactivate your domain by changing the mode of your valid user table from "Defer All" back to the appropriate mode.

The error message that caused a bounce was:
bounced (host WAN1.yourdomain.com[70.xx.xxx.253] said: 550 5.1.1 User unknown (in reply to RCPT TO command))

This is being done in an effort to eliminate backscatter (or outscatter) from our servers. When using our mail services, our servers should be trusted or whitelisted by your mail server. If you need to blacklist IP addresses, recipients, or perform any filtering tasks, please use the tools in the account control center. If a filtering tool you would like to use is not in our services, let us know; we are happy to add new filtering options. A properly configured valid user table should also be present for each mail domain.
---------------

I had originally set rollernet to allow all mode, but this somehow got changed during the outage/downtime it would seem..

The only option that has worked since then was setting their service to use a global table..

However, at one point i had the server offline after it was working, did a few test messages, but none of those messages showed up in rollernets queue, though once i turned things back on they did show up in the inboxes of our exch. server.

I even tried adding users to the global table on rollernet, but this didnt result in seeing email getting queued there either. (when i did these tests i had the mx records set to the wan1 10, wan2 10, rollernet 15, etc setting, not the original setting during the original failure).

Has anyone worked with rollernet.. Or.. better yet.. can anyone recommend an inexpensive alternative backup mx solution out there?

Any assistance or thoughts on this would be great...

I have to explain to upper management why end users in the outside world didnt even get a bounce message, let alone noone locally getting the messages sent from outside.

Thanks

 

[crymedry]

Was there ever an answer to this question? I am having a similar issue and can't figure it out.

I have just installed a new exchange server 2003 and am migrating from a third party pop provider to a new exchange server. I have setup mx records and everthing looks right. However, I can send email just fine, but when I reply to those emails or try to send a new one, I don't get a bounce back nor does it show up in my inbox .... what am I missing?

 

[markm75]

I never really had an answer to mine.. however, I think i've narrowed ours down to this..

We had hardware failure.. we have a backup mx service provider in rollernet.us.. ie: we have our regular mx record, then the rollernet.us addresses as well (with lower priorities)..

For some reason, we werent notified of a change in service from them.. they no longer support the "allow all"type mode setup.. you have to configure ridiculous user tables for all of your users, otherwise the emails get bounced to prevent what they call backscattering (spammers)..

By the time I realized the rollernet account was setup as such it was too late.. we lost almost 2 days worth of incoming emails. They had a log of the senders.. which I tried to access yesterday, I had the list on my screen, closed the window, went back a few hours later to cut and paste the list and their database from the date in question was deleted (i guess i waited too long, but didnt realize this was possible till yesterday, to view the logs).

So now we are going to go with another backup mx service provider (the regular ISP only holds mail for 48 hours if that, they dont offer their own service and there is no logging).

I was either going with dnsexit.com ($19 a year) or Dyndns $29.99 a year. Rollernet was free, but with their new restrictions, its not as attractive, plus their lack of notification on changes was a very bad business thing.

---

Crymedry.. in your case you said you migrated from a 3rd party pop provider and that outgoing messages are fine.. but attempts to reply to those messages on the outside fails without warning?

Have you verified your smtp settings in the Exchange snapin? Often times, incoming communication issues start here...

IE: There is a setting to specify your mail server dns name in there.. ie: server.mydomain.com where this same address usually appears as the mx record in DNS as well.

Is your DNS settings held on the outside world or within your AD environment for your external domain (email server)..

 

[markm75]

Has anyone ever run into a backup mx service that allows you to customize the bounce back period.. ie: change the 10 days to say 2 days?

 

[bustamove]

Crymedry, How did you change your MX record? Did you do the following?
1. check your orignial TTL for MX, and write it down, say it is 86400,
2. change TTL to something really short, say 300, and wait for more than 24hrs.
3. change the MX record

also make sure your ISP dont do mail caching