From owner-freebsd-hackers Mon Feb 5 14:12:47 2001 Delivered-To: freebsd-hackers@freebsd.org Received: from gandalf.vi.bravenet.com (gandalf.bravenet.com [139.142.105.50]) by hub.freebsd.org (Postfix) with SMTP id 5E08E37B491 for ; Mon, 5 Feb 2001 14:12:28 -0800 (PST) Received: (qmail 23628 invoked by uid 1000); 5 Feb 2001 22:12:00 -0000 Received: from localhost (sendmail-bs@127.0.0.1) by localhost with SMTP; 5 Feb 2001 22:12:00 -0000 Date: Mon, 5 Feb 2001 14:12:00 -0800 (PST) From: Dan Phoenix To: Alfred Perlstein Cc: scanner@jurai.net, freebsd-hackers@FreeBSD.ORG, freebsd-question@FreeBSD.ORG Subject: Re: qmail IO problems In-Reply-To: <20010205135501.H26076@fw.wintelcom.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Actually another point i should bring up is there are 4 separate perl scripts that basically grab users from a mysql database.... their mail message is taken and piped to...sendmail -t which is a symbolic link to /var/qmail/bin/sendmail -t ...someone told me about a concept of "herding the queue" doing it that fashion....I am quite interested if there is a more effective way of doing mail with qmail and perl...maybe the way ezmlm does it so quickly. But the I/O problem is still there without perl scripts running from all 30 internal fbsd webservers. On Mon, 5 Feb 2001, Alfred Perlstein wrote: > Date: Mon, 5 Feb 2001 13:55:01 -0800 > From: Alfred Perlstein > To: Dan Phoenix > Cc: scanner@jurai.net, freebsd-hackers@FreeBSD.ORG, > freebsd-question@FreeBSD.ORG > Subject: Re: qmail IO problems > > * Dan Phoenix [010205 13:50] wrote: > > /0 /1 /2 /3 /4 /5 /6 /7 /8 /9 /10 > > Load Average |||| > > > > /0 /10 /20 /30 /40 /50 /60 /70 /80 /90 /100 > > cpu user|X > > nice| > > system|XXXXXX > > interrupt| > > idle|XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX > > > > /0 /10 /20 /30 /40 /50 /60 /70 /80 /90 /100 > > ad0 MB/s > > tps|XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX108.83 > > > > > > that is why i know it is IO....#1. > > #2..not configuration errors that I know of. > > tailing the tcpserver current log shows lots of traffic comming in and > > out.....yet it keep getting queued up......so my guess is the IO. > > Post fix is not the solution...would be the exact same problem...a mail > > deamon was to read each message to send them right? Same disk I/O problem > > will occur. Btw who are you and when you say brad....who are you talking > > about? > > Here's a couple of tunables you might want to try, use then one at a > time and see if it improves things: > > vfs.vmiodirenable = 1 > vfs.write_behind = 0 or 2 (default is 1) > > You might also want to consider a more robust disk subsystem, using > a single IDE non-redundant disk for storing mail isn't such a hot > idea. > > -- > -Alfred Perlstein - [bright@wintelcom.net|alfred@freebsd.org] > "I have the heart of a child; I keep it in a jar on my desk." > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message