I am setting up a server to perform emails to our users, so I use all possible ways to minimize spam problems.

Yandex has an FBL notification system (Feedback Loop) , which sends to the specified email address reports about users adding spam emails that came from a specific network (networks), so that later on these reports, the sender removes the addresses of these users from his distribution list ( again do not send letters to those who add them to spam).

My problem is that there is nothing in these reports that could identify the letter and the sender. Apparently, Yandex (or returnpath.net, judging by the headings of the reports) hides the addresses of users by writing noodles like ca2b45 ... @ email.xxx ([32-character hash-like-md5] @ email.xxx) instead , including the To and Original-Rcpt-To headers, and also does not transmit anything from the original message, except for the subject (Subject).

The FAQ page on YandexFBL has the question “How should I process complaints from the FBL report?”, In response to which it is written:

"The ISP can edit parts of the original messages to protect the personal data of their users. If the ISP has deleted the subscribers' addresses from the FBL report, you should include the control IDs that are unique for the recipients in your database and match those messages were sent to the addressee to track the deletions. You can add unique recipient IDs to the message body, or create a special "X" header: for example, "X-SenderName-ClientID: 543432". "

But the problem is that the body of the original message is not forwarded, as are the X-SenderName-ClientID / Message-ID / X-Feedback-ID / Feedback-ID headers. All that can be recognized by the sent letter, which was added by spam is its topic, but I do not think that adding any identifiers to the subject of each message is a good idea.

For example, when working with Mail.ru FBL, everything happens much simpler, since in their reports the user's address is not replaced, and the original Message-ID is also transmitted.

Who faced a problem, please tell me how to add a message identifier to reports from Yandex FBL

    2 answers 2

    After some searching, I found the answer:

    On the Yandex Post Office - Troubleshooting page, in the form of an informational message at the very top of the page (not in the list of questions, as expected) it says: "FBL report. Message-ID and another links. We’ve been using the Yandex.Mail users. "

    In other words, Yandex deliberately tore out of the reports of the address and any data that can identify the letter, leaving only the information that "Someone put the message with such a topic in spam". I suppose their only use now is to count those very additions to spam, and in the case of the massive receipt of such reports - a signal to action ("Someone sends spam from your network with this topic, EOF").

    In short, the answer to the question: "The report ID cannot be added to the report (in the normal way)."

    You can add an identifier to the message header, as it comes in the report, but this can negatively affect mailings (if I'm not mistaken, many require that the mailing topics for all addressees be the same).

    A few more conclusions - outdated data are posted on the Yandex FBL FAQ page, and it will not be possible to automate the processing of messages about spam from Yandex.

      Yes, the author of the question (answer) is absolutely right. Yandex in terms of providing information in the FBL report is the most ... of all the other major mail services that provide reports.

      1. Using additional information in the subject of the letter to identify the addressee (and this is the only field that Yandex does not modify in its report), as noted above, will lead to negative consequences. It is necessary to adhere to the rule: one mailing is the same subject of letters.
      2. Essentially. I was able to resolve this issue only with the help of the letter id which prescribes the mail server (Exim) in the headers and Yandex does not touch it. But there are several inconveniences (and very global ones in the experience of using this algorithm):
        • You need to parse the logs of the mail server and extract data from there - a pair: Email recipient and letter id.
        • need to store this data somewhere.

      Unfortunately, this is an additional load on server resources, and I have not yet found another option.