[Mimedefang] Debugging Signal 14 - slave dying unexpectedly

Matt Smith Matt at coolchilli.com
Wed Aug 8 23:35:42 EDT 2007


Hi All,

Recently (about a month it seems) one of my MD installations has been 
having slaves die unexpectedly with the above return signal.
This happens on consistent messages, however mail IS flowing through 
normally otherwise for Spam, viral and ham. Only some messages are 
failing, and they fail consistently.

Viewing the archives, it seems this is related to *something* in 
SpamAssassin, and adding my own debug lines into the filter seems to 
confirm this (nothing is seen after the spam_assassin_check() call).

I have been successful in quarantining messages that match the previous 
criteria (sender/recipient), however I have *NO* idea if these are the 
same messages that failed previously, as I cannot quarantine those due 
to the slave dying before that takes place (kind of a chicken/egg 
scenario!).

I've run (as defang) spamassassin -D -p /etc/mail/sa-mimedefang.p 
INPUTMSG against the quarantined messages and don't see any critical 
errors/strange output.
I've also disabled bayes to see if that was doing it, but still the 
errors persist.

 From my reading (of the archives), something in SpamAssassin is setting 
SIGALRM, and not clearing it, how can I find this?
Where/how else can I track down what is causing these slaves to die?


Thanks for ANY assistance, this has been my morning, and looks to be the 
remainder of my day!

Cheers,
Matt



More information about the MIMEDefang mailing list