Yesterday we started absorbing mydoom instead of bouncing them. In around 24 hours, our main MX server for perl.org/cpan.org absorbed 1.3 GB of mydoom. That's 41083 messages. Or, 1710 an hour. Or, 29 a minute. Or, about one Mydoom every 2 seconds. Ouch!
In addition, we absorbed 354 MB of bounces, in 17394 messages. (One bounce every 5 seconds or so.)
The good news is that things seem to have stabilized. We have the machines configured such that they are no longer pegged to the wall. Occasionally a burst of mail will make them lean a little, but they're handling just fine. Mail is moving at full speed, transiting our system in seconds, instead of minutes or hours.
I hope the flow of mail slows soon - even though we are putting plans in place for more servers - I miss the old days, where you could just run sendmail and have it work.
2000 mydoom an hour, 2500 bounces an hour (Score:3, Interesting)
Reply to This
configuration changes (Score:2, Interesting)
And, if you've managed a 100% solution, a config file diff might also go a long way towards helping other admins (who also are not mailserver gurus
Re:configuration changes (Score:2, Informative)
Re:configuration changes (Score:3, Interesting)
Re:configuration changes (Score:4, Interesting)
We're using a hacked up qpsmtpd [develooper.com] (soon to be rolled back into the main dist). Basically, we have a rule that matches mydoom, and directs it to a folder.
The rule is overly liberal right now - getting hit with 2 of these a second, spamassassin and clamav would grind us to a halt. But that's ok.
Of the 25000 emails caught in my mydoom filter in the past 13 hours, there are 1322 unique (case insensitive) subject lines. None of them look like anything someone will miss.
Reply to This
Parent