On 10 Jan 2006 at 12:09, David Phillips <[log in to unmask]> wrote: > With mail-merge on: > 1. mail is going out just fine to nearly every subscriber. > 2. One ISP, verizon.net, is bouncing these emails. We get a relaying > not allowed message, then a bounced message in our postfix logs. > Verizon.net is the only ISP that's bouncing right now. > > With mail-merge off: > 1. We have no problems with sending email to any of our subscribers > 2. Verizon gives us no problems. > > The only difference we've been able to spot is the difference in the > Return-Path header field. > > We do NOT know, however, if that difference is the actual cause of > verizon refusing us. If it is, then it's a misconfiguration on Verizon's part. HOWEVER, I don't think so because my personal email address is at verizon.net and I just tested it. I receive mail-merge list mail there just fine, with the probe return address and everything. > The errors we see in our postfix log, when mail-merge is on, are: > Jan 7 04:01:06 shelmerston postfix/smtp[3952]: E6B2C73E73: > to=<[redacted]@VERIZON.NET>, relay=relay.verizon.net[206.46.232.11], > delay=50, status=bounced (host relay.verizon.net[206.46.232.11] said: > 550 You are not allowed to send mail:sv6pub.verizon.net (in reply to > MAIL FROM command)) > > Jan 7 04:01:06 shelmerston postfix/local[3930]: 68C7D63F66: > to=<owner-gunroom*[redacted]**VERIZON*[log in to unmask]>, > relay=local, delay=0, status=bounced (unknown user: > "owner-gunroom*[redacted]**verizon*-net") > > Is there a reason that the Return-Path has to be changed when > mail-merge is on (that's what I don't understand yet.) The only reason NOT to use probes is to save bandwidth by sending bulk email. If you're going to use mail-merge, you might as well use probes so that bounce processing works 100% of the time (assuming the receiving MTA sends to bounce to the right place) instead of whatever the percentage is on non-probe bounces. Is there a reason it HAS to be that way? Well, from a technical point of view, it would be _possible_ for us to implement mail-merge such that it does not use probe bounces, in the same way as it would be possible for a car manufacturer to build a car without a seat belt -- but would it really make sense to do so in this day and age? -- Francoise Becker <[log in to unmask]> Knowledge is just a click away: http://www.lsoft.com/optin.html