We have set it so that the subscribers are set to NOMAIL, if a total error count is 2.

In your case, if you select a week (not the same as 7 days), the error counter is cleared on a Sunday (according to the help).

Your initial period could have been shorter than a week, or if your e-mail server was down when going from Saturday to Sunday, then I guess the clearing would not take place.

Another guess is that the error handling routines, might not be in synch with the user-statistics, as you change mode. F.ex. if you edit the lists.mer file manually you have to restart Mercury.

But this is guesswork from my side.

Hello,

We use Mercury as a list server. I had two active lists and I turned on VERP processing yesterday.

My verp settings were this...

"errors allowed before handling occurs" = 1

"delete unsubscribe the address from the list" = yes

"mail a summary of verb changes to moderators" = daily

I received the daily verp changes summary and it contained about 1500 addresses that were unsubscribed.

But we noticed an issue in it. There were addresses in the unsubscribe that I know for a fact were valid/active addresses but for some reason verp unsubscribed them. An example is yyyyy@eeeee.com which I know is a valid address but I also know that the remote e-mail system send e-mail as yyyyy@eeeee-ffffff.com but receives e-mail with just the english domain name yyyyy@eeeee.com . So I did some testing with a test-list I put yyyyy@eeeee.com as a subscriber and set the verp settings to the same as a production list and sent a test msg to the list. I know the msg was received by yyyyy@eeeee.com ok and the verp listing did not remove them for the test-list.

It shoud be noted that in the production list verp summary there were no addresses unsubscribed who used the bilingual domain name such as yyyyy@eeeee-fffff.com only yyyyy@eeeee.com addresses were removed from the production list (that were valid)

Does anyone have an idea as to what caused this issue ?

Under what exact conditions does verp consider unsubscring someone as valid ? "mail box does not exist" ? "mailbox full" ? what else ?

 Any suggestions greatly appreciated.

Let me know if you need more information.

We are running Mercury 4.72 July 10,2009 on Windows XP fully patched

 Thank you

Neil

P.S. Is the mercury@bama.ua.edu e-mail support list down ? I sent two msgs to it last night and never saw anything.

Increase the error count, and set only to disable not to delete as a normal precaution.

We have error count set to 3 so that graylisting implementations do not trigger this.

We've found that any smtp return code that is not succesful triggers the verp error count, even if the message is delivered a little later - so you need a little more tolerance there. This also means that a server that no longer serves a maildrop in your subscription list, might take a few messages before the subscriber is unsubscribed/disabled.

Thank you for the reply Peter.

I basically did what you suggested yesterday. I set the error count to 3 in a 7 day period.

Today one msg was sent to the list in question and tonight I got a verp process listing and it contained 4 subscribers that were unsubscribed.

My question is why were they unsubscribed in one day with only one msg being sent to the list. Should they not be unsubscribed only after 7 days ?

Or am I missing something here? Now that I think of it it means 3 errors not 3 msgs sent to the list but given this why did it happen in 1 day and not 7 ?

Thank you

Neil