This post is mostly to help me remember how to do this, if the situation arises again.

I just had a lot of mail backup on my server. The 10MinuteMail inbox was over 300 MB (usually it kept below a megabyte), Postfix’s active queue was maxed out at 20,003 entries (why the 3, I don’t know), and the incoming queue was another 20,000+. Basically everything was all backed up. I’m not 100% sure how this condition gets started. I’ve seen it a few times on my old server when super high volumes of incoming mail deliveries combined with other sites I hope serving up high bandwidth to end users. This is the first time it’s happened on the new server. It may be time to change out the domain that the 10MinuteMail e-mail addresses are using.

Regardless, using qshape I was able to identify a handful of from addresses (presumably either spammers or a cyclic bounce issue) which accounted for over 8,000 of the mail in the active queue. By using the following command I was able to purge out just those messages from the queue:

mailq|awk ' /^[0-9A-F][0-9A-F]*.*error.mag2.com$/ {print $1}'|tr -d '*'| xargs -rn1 postsuper -d

Where error.mag2.com is the domain, or from address you wish to delete. This works pretty well. I may whip up a bash script to handle this in the future.

For reference, the worst offenders are:

  1. magerr.combzmail.jp
  2. prjapanmail.jp
  3. error.mag2.com
  4. accessmail.jp
  5. mayld.net

Why so many from Japan? I have no idea….