DOOM: High CPU and timeouts with "spamd child" and spamass-milter

So, in my ongoing series on Weird Problems Google Didn't Help With I give you the following problem.

Our trusty mail server has Postfix passing mail to spamass-milter. Alas, something goes wrong. Although logs show spamd returning a result postfix complains about connections timing out.

Logs point to a SMFIC_BODYEOB error. What the?

/var/spool/mail.log:

Jun 27 19:26:02 myhost spamd[10058]: spamd: clean message (-2.9/5.0) for spamass-milter:117 in 7.1 seconds, 959 bytes. 
Jun 27 19:26:02 myhost spamd[10058]: spamd: result: . -2 - ALL_TRUSTED,BAYES_00,TVD_SPACE_RATIO scantime=7.1,size=959,user=spamass-milter,uid=117,required_score=5.0,rhost=localhost,raddr=127.0.0.1,rport=42711,mid=,bayes=0.000000,autolearn=ham 
Jun 27 19:30:55 myhost postfix/cleanup[15798]: warning: milter unix:/spamass/spamass.sock: can't read SMFIC_BODYEOB reply packet header: Connection timed out 
Jun 27 19:30:55 myhost postfix/qmgr[2757]: 8CD3F58: from=, size=1100, nrcpt=1 (queue active) Jun 27 19:30:55 myhost postfix/smtpd[15795]: disconnect from unknown[192.168.1.6] 


In my case it turns out that I'd made the mistake of having two OPTIONS lines in my /etc/default/spamass-milter file.

OPTIONS="-u spamass-milter -i 127.0.0.1"

# Reject emails with spamassassin scores > 15.
#OPTIONS="-r 15"
OPTIONS="-r 15"

Having the two OPTIONS lines is fine, as long as you only want the last option - in this case not telling spamass-milter what default user to use or that it shouldn't check things originating from the local machine.

I'm not entirely sure why it causes spamd child to go haywire and eat up all the CPU for five minutes, but I'm glad it's stopped.

Comments

Post a Comment

Popular posts from this blog

Troubleshooting Windows Server VSS issues without rebooting

After being told by our backup crew that I'd probably have to reboot any Windows servers with VSS issues I had a look through all the information I could find. Alright, I should have read about this in 2003 odd, but it's not really my field of IT expertise. Having collated and written up some notes for work it seemed a shame to not post them here. Contents Volume Shadow copy Service component overview Overiew of VSS troubleshooting  Resolve   Retry the operation and restart the writer if necessary   Retry the operation and restart the writer   Open Event Viewer and view events related to VSS  Verify Resetting VSS Clients/Requestors  Resetting VSS Writers  Another random page Deleting Shadows Resetting Providers  Microsoft Software Shadow Copy Provider Tools  VSSAdmin  VSSUIRUN  Powershell Get-VSSWriters Other reading Volume Shadow copy Service component overview Writers: VSS aware add-ons that handle writing consi...
Read more

Cage Match - Gluten vs Holy Communion

I passed a local church yesterday and was quite surprised to see a their huge motivational board proudly proclaiming "ALLERGY / GLUTEN FREE COMMUNION BREAD". That got me thinking... a dangerous activity while driving. If communion magically transforms communion nibbles into the actual body of a Christ, wouldn't it be gluten free anyway. Today Google/Wikipedia sucked me down and into the rabbit hole that's the swirling vortex of Christian dogma. If I'm going to have my brain explode into confused lumps I want to share - so here we are. What we have here is the Catholic tradition of the Eucharist, followed to varying degrees in other Christian sects. The wafer and wine are blessed and undergo Transubstantiation - transforming into the body and blood of Jesus. It turns out that it's not as simple as the communion wafer actually turning into the physical body of Christ. I mean, you can test that sort of thing. The wafers would go all floppy - unless it's ...
Read more

Revenge of SMFIC_BODYEOB reply packet header

My old friend SMFIC_BODYEOB  put in another appearance today. Or, more probably, the last six months and I haven't noticed. Jan 12 16:41:20 myserver postfix/cleanup[4860]: warning: milter unix:/spamass/spamass.sock: can't read SMFIC_BODYEOB reply packet header: Connection timed out GAH! I thought I'd fixed this. Off to the /etc/default/spamass-milter I ran. No obvious problem... Next, head scratching. Then, comparing to my test server, I added in "-d spamc" to the  OPTIONS line - which shows input/output to spamc in the mail.log Sure enough - although spamd was returning a result in about ten seconds (not especially speedy, but it is a Pi) spamc was taking five minutes to return those results to spamass-milter. By then Postfix had given up and delivered the mail. Jan 12 22:46:03 myserver postfix/smtpd[3108]: connect from localhost[127.0.0.1] Jan 12 22:46:03 myserver spamass-milter[2979]: output "X-Envelope-From: <ldhsnvf@ca-se-visite.fr>#0...
Read more