Which MTA are you using? Exim? Well I had the same problem last month and everything got fixed when I limited the message scan size to 80k.
SA was sacanning everything even 10mb messages so I reduced the scan to 80K by adding a rule to exim.conf. Now I dont see spikes at all. Load is around 3 now and spamd child process are 30mb in size each one. If anyone suggest to remove some SA plugin from v310.pre, v312.pre or v320.pre would be nice. Maybe one of this plugins is pushing the load too, If anyone recommend to turn it off, I would like to hear comments: Perform DCC checks Perform Pyzor checks Perform Razor2 checks Perform Spamcop reporting Perform AWL checks AutoLearnThreshold WhiteListSubject MimeHeader ReplaceTags URIDNSBL HashCash SPF ImageInfo VBounce Bayes BodyEval DnsEval HTMLEval HeaderEval MIMEEval RelayEval URIEval WLBLEval URIDetail HTTPSMismatch Check Webmaster-163 wrote: > > Spamassassin is using a lot of resources, in particular, > high CPU for spamd running as one of my users. I am > now seeing spam leak through. I have replaced the spamd > option '-m5' with '--round-robin' in my start up script to > no effect. Other users demand for spamd appears normal. > I am using RulesDuJour and SARE rulesets. What would > cause this problem and what can I do about it? Thank you > in advance! > > setup: SA 3.2.2 w. procmail & Sendmail on fedora core 5 > > -- View this message in context: http://www.nabble.com/spamassassin-gets-stucked-in-tf4206163.html#a12051965 Sent from the SpamAssassin - Users mailing list archive at Nabble.com.
