Single process taking over?

[Julian Field]

Can you try switching off Bayes (use_bayes 0 in spam.assassin.prefs.conf).
Then let me know if the problem recurs.

Also, upgrade your SA to 2.63 in case you are seeing a bug in SA.

At 15:31 17/02/2004, you wrote:
>Mr. Field,
>
>I have been going through pretty much the same situation as described with
>this post. The exception is that my machine does not show a domineering
>process and load average drops to near nothing. I have been trying to change
>sendmail to remedy this problem, but I may be looking at the wrong part of
>the puzzle. I have still not determined what is going on, but I do see a lot
>of Bayes lock files and one main bayes.lock file. It peaks once I see the
>bayes_toks.new file which seems to stay around forever. I offer this only to
>maybe point things toward a solution.
>
>I am running
>     MS 4.26.8-1
>     SA 2.61-1
>     ClamAV 0.65
>     MailWatch 0.5.1
>     Sendmail 8.11.6-27.73
>     RH 7.3
>
>I upgraded MS on a Monday and MailWatch on a Wednesday. That week problems
>started happening. The problems seemed to be resolved by my removing my
>Bayes files (you suggested poisoning, and this appeared to have been the
>case), but since I must stop MS, remove all of the bayes lock files, the
>bayes_tok.new file, restart MS and all appears fine. Load average climbs to
>normal to normal-high limits, my incoming backlog clears quickly and
>everything is fine. I replaced my Message.pm file with the the one you
>posted to the list, and that is the only other change I have made to the
>above installed programs.
>
>I hope some common thread may appear from my configuration and what others
>describe to shed some light on this. Most people don't complain about load
>averages this low, but to me it signals a slow down in my mail system,
>creating backlogs in the incoming queue.
>
>Thank you for your efforts, sir.
>
>Steve Campbell
>campbell at cnpapers.com
>Charleston Newspapers
>
>----- Original Message -----
>From: "Julian Field" <mailscanner at ECS.SOTON.AC.UK>
>To: <MAILSCANNER at JISCMAIL.AC.UK>
>Sent: Tuesday, February 17, 2004 7:24 AM
>Subject: Re: Single process taking over?
>
>
> > At 11:38 17/02/2004, you wrote:
> > > > -----Original Message-----
> > > > From: MailScanner mailing list [mailto:MAILSCANNER at JISCMAIL.AC.UK] On
> > > > Behalf Of Julian Field
> > > > Sent: Tuesday, 17 February 2004 8:36 PM
> > > > To: MAILSCANNER at JISCMAIL.AC.UK
> > > > Subject: Re: Single process taking over?
> > > >
> > > > Ah, a reproducible fault! I like those :-)
> > >
> > >I don't!!  :)
> > >
> > > > What does your MailScanner.conf look like? (just the interesting bits,
> > > > don't care what all the filenames of your reports are and stuff like
> > > > that).
> > >
> > >See below..
> > >
> > > > What virus scanner(s), SpamAssassin, etc?
> > >
> > >ClamAV, McAfee, Spamassassin 2.6.3, DCC, Razor
> > >
> > > > What is the last thing the runaway process logs before CPU hogging?
> > >
> > >Nothing abnormal, just the process starting and mail being processed,
>even
> > >in verbose logging, it just appears to be a normal process that won't let
> > >the other threads have any resources.  If I kill it, the other threads
>spawn
> > >and run as per normal.
> > >
> > > > Does the CPU hogging start the instant you start MailScanner, or the
> > > > instant the first child process runs, or when?
> > >
> > >As soon as mail begins to be processed.  If no mail is in the queue, not
> > >hogging, but the second any mail is in queue it's hogging.
> > >
> > >MailScanner.conf
> > >============================================
> > >Max Children = 4
> > >Queue Scan Interval = 1
> > >MTA = sendmail
> > >Max Unscanned Bytes Per Scan = 100000000
> > >Max Unsafe Bytes Per Scan = 50000000
> > >Max Unscanned Messages Per Scan = 15
> > >Max Unsafe Messages Per Scan = 15
> > >Virus Scanning = yes
> > >Virus Scanners = mcafee clamav
> > >Virus Scanner Timeout = 300
> > >Spam Checks = yes
> > >Spam List =
> > >Use SpamAssassin = yes
> > >Max SpamAssassin Size = 90000
> > >Deliver In Background = yes
> > >Delivery Method = batch
> >
> > Do you reckon you could reproduce the problem on a box to which you could
> > give me login access? I suspect it's something very simple, but I have
> > never witnessed it here and it's apparently not a common problem.
> > --
> > Julian Field
> > www.MailScanner.info
> > MailScanner thanks transtec Computers for their support
> >
> > PGP footprint: EE81 D763 3DB0 0BFD E1DC 7222 11F6 5947 1415 B654

--
Julian Field
www.MailScanner.info
MailScanner thanks transtec Computers for their support

PGP footprint: EE81 D763 3DB0 0BFD E1DC 7222 11F6 5947 1415 B654