Installing big-concurrency.patch
Hi, What's the correct procedure for installing the big-concurrency patch on a live system? Is it just a patch, make, make setup check, restart qmail? Or will it affect/break anything else? Thanks, Brian -- Brian Baquiran <[EMAIL PROTECTED]> http://www.baquiran.com/ AIM: bbaquiran Work: (632)718 Home: (632)9227123
qmail initiating identd lookups?
Is there any component of qmail/ucspi-tcp that would initiate identd lookups on a remote host? A network admin has complained that his server has been receiving a large amount of identd queries to his server coming from our mailserver. Here's the excerpt from the logfile he sent us. 210.16.71.2 is our mailserver. Apr 6 07:40:11 lillie /kernel: ipfw: 500 Deny TCP 210.110.247.244:53 161.184.159.82:111 in via ed0 Apr 6 08:06:37 lillie /kernel: ipfw: 500 Deny TCP 210.16.71.2:4963 161.184.159.82:113 in via ed0 Apr 6 08:06:58 lillie last message repeated 3 times Apr 6 08:36:04 lillie /kernel: ipfw: 500 Deny TCP 210.16.71.2:3168 161.184.159.82:113 in via ed0 Apr 6 08:36:22 lillie last message repeated 2 times Apr 6 09:06:12 lillie /kernel: ipfw: 500 Deny TCP 210.16.71.2:1153 161.184.159.82:113 in via ed0 Apr 6 09:06:33 lillie last message repeated 3 times Apr 6 10:05:51 lillie /kernel: ipfw: 500 Deny TCP 210.16.71.2:3246 161.184.159.82:113 in via ed0 Apr 6 10:06:12 lillie last message repeated 2 times Apr 6 10:39:18 lillie /kernel: ipfw: 500 Deny TCP 210.16.71.2:4250 161.184.159.82:113 in via ed0 Apr 6 10:39:36 lillie last message repeated 2 times Apr 6 11:06:09 lillie /kernel: ipfw: 500 Deny TCP 210.16.71.2:4038 161.184.159.82:113 in via ed0 Apr 6 11:06:18 lillie last message repeated 2 times Apr 6 12:07:37 lillie /kernel: ipfw: 500 Deny TCP 210.16.71.2:1681 161.184.159.82:113 in via ed0 Apr 6 12:07:46 lillie last message repeated 2 times Apr 6 12:35:40 lillie /kernel: ipfw: 500 Deny TCP 210.16.71.2:1410 161.184.159.82:113 in via ed0 Apr 6 12:36:01 lillie last message repeated 3 times Apr 6 13:05:38 lillie /kernel: ipfw: 500 Deny TCP 210.16.71.2:1899 161.184.159.82:113 in via ed0 TIA, Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: Still problem with yahoo.com
On Tue, 4 Apr 2000 [EMAIL PROTECTED] wrote: > On Tue, Apr 04, 2000 at 10:47:59PM -0400, Bryan White wrote: > > > > > > > Neither of yahoo.com's mail exchangers is answering right now. It's not > > your > > > problem. Don't worry about it. > > > > Any idea how long till they will be back up? Is there someplace to find > > this information. I have been struggling with mails queues clogged with > > messages for yahoo.com all day. I am on the virge of rejecting new messages > > to the yahoo domain. > > In what way are you "struggling"? Sure there may be plenty of emails in your queue, > but is it really hurting? I can't speak for anyone else, but In my case, yes I am hurting. We maxed at 10k queued messages this morning, and it's eating up the concurrencyremote. Mail that's NOT for yahoo.com also gets stuck for several hours as well. > Or do you just not like that idea of qmail-remote failing > and seeing the corresponding log entry? Remote delivery is very cheap with qmail so > you don't need to be too concerned unless the yahoo mail is consuming all of your > concurrencyremote. If it's not, then you can sleep soundly as it's no big deal. Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: tcpserver fork
On Fri, 31 Mar 2000, Peter van Dijk wrote: > On Fri, Mar 31, 2000 at 01:06:10PM +, S Ashok Kumar wrote: > > RedHat 6.1 with qmail-1.03 and ucspi - 0.84. > > qmail-smtpd is run by tcpserver and the -c value is 2000. But as soon as > > it reaches 1247 sessions, it is not able to fork further processes, the > > error message is: > > > > tcpserver: warning: dropping connection, unable to fork: temporary > > failure > > > > You are out of processes. A little tweaking with 'ulimit' might help, or > perhaps you need to tune your kernel somewhere deeper. For Linux, you need to increase NR_TASKS. Edit your /usr/src/linux/includes/linux/tasks.h and recompile the kernel. I've already bumped up NR_TASKS to 1024, then again to 2048. From the comments in tasks.h, the limit for NR_TASKS is 4096 on intel, but I haven't been able to get it to boot when set to 4096 (this is on 2.2.14, IIRC, so newer/development kernels may be different.) HTH, Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: Running qmail on a 4x Xeon 550MHz system
On Sat, 25 Mar 2000, Adam McKenna wrote: > On Sat, Mar 25, 2000 at 11:36:44PM +0800, [EMAIL PROTECTED] wrote: > > I agree. I manage a similar system (quad Xeon, 1GB memory) with a > > loadaverage of 0.3 or thereabouts. > > In other words, your quad Xeon is being wasted. You could replace it with a > single Pentium II or Pentium III machine. Depends on what you mean by wasted. Nobody's worried about the mail system being overloaded (bandwidth now becoming the limiting factor). It keeps The Man off my back. IMHO, a small price to pay. Of course, yes, it could be handled by a PII for now, but I'd still rather have the quad Xeon with the redundant hardware goodies (and a service contract) than an off-the-shelf PII, or even one built by my own hands. > > Although I'm not familiar with the intricacies of how load is > > distributed across multiple CPUs among different OS's, I would suspect > > something is wrong with your system. > > That is an absolutely terrible thing to say. You don't know anything about > his system Granted, I'm not a sysadmin by profession, and I've only been using qmail for ~4 months, but I've been running a similar system (aside from the ReiserFS) under a similar load. From my limited experience, and our record of loadaverages over the past 3.5 weeks, a load average of 1.4 would indicate something out of the ordinary. > (and obviously, not enough about Unix or computers in general) to > make an accurate analysis based on his post. Please refrain from saying such > things in the future unless you're sure you know what you're talking about. Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: Running qmail on a 4x Xeon 550MHz system
On Sat, 25 Mar 2000, Ruben van der Leij wrote: > On Sat, Mar 25, 2000 at 12:48:04PM +, Andreas Aardal Hanssen wrote: > > > Does anyone have any idea what could be the problem? The > > disk IO is very low and my computer is *really* sleeping, > > with a load average (uptime etc) of approx. 1.4.. > > A loadaverage of 1.4 means you have on average 1.4 task waiting to run. Or, > to put it in percentages: your machine has 140% of it's time filled with > tasks that want to run. > Even for an idle quad Xeon, that is way to high. I agree. I manage a similar system (quad Xeon, 1GB memory) with a loadaverage of 0.3 or thereabouts. Although I'm not familiar with the intricacies of how load is distributed across multiple CPUs among different OS's, I would suspect something is wrong with your system. Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
qmail-qsanity error messages
I just ran qmail-qsanity on a mailserver. It reports: message has no entry in info: 492326 message is neither local nor remote: 492326 message has no entry in info: 492165 message is neither local nor remote: 492165 message has no entry in info: 492166 message is neither local nor remote: 492166 How do I fix these messages? Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: tcpserver unable to fork?
On Fri, 17 Mar 2000 [EMAIL PROTECTED] wrote: > Today I got the following message in my SMTP logs: > > tcpserver: warning: dropping connection, unable to fork: temporary failure > tcpserver: status: 146/256 In an act of desperation, I started another tcpserver to listen on another interface. I noticed that I can only run a max of around 150 smtpd's simultaneously, whether under one tcpserver or two. > exec /usr/local/bin/softlimit -m 200 \ > /usr/local/bin/tcpserver -c256 -v -p -x /etc/tcp.smtp.cdb \ > -u $QMAILDUID -g $NOFILESGID 0 smtp \ > /usr/local/bin/rblsmtpd /var/qmail/bin/qmail-smtpd 2>& > > Can someone explain what's happening? Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
tcpserver unable to fork?
Today I got the following message in my SMTP logs: tcpserver: warning: dropping connection, unable to fork: temporary failure tcpserver: status: 146/256 I'm currently running qmail-smtpd under supervise with the following line in my run file, broken for readability: exec /usr/local/bin/softlimit -m 200 \ /usr/local/bin/tcpserver -c256 -v -p -x /etc/tcp.smtp.cdb \ -u $QMAILDUID -g $NOFILESGID 0 smtp \ /usr/local/bin/rblsmtpd /var/qmail/bin/qmail-smtpd 2>& Can someone explain what's happening? Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: Virtual Domains
On Tue, 14 Mar 2000, MicroSense Computer Works wrote: > I am currently running Paul Greggs method of working with virtual domains. Works >Great.=20 > > The only problem that I have with it is administration. I don't want to write my >own user / Domain manager to add and delete accounts (nor do I want to do it manually) > > Any good suggestions on an easy to use virtual domain manager. I happen to like vpopmail (http://www.inter7.com/vpopmail), but then again I'm only managing one large domain. It has optional MySQL support, and has commands to add/delete users and domains, change passwords etc., and has support for user quotas. HTH, Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: How big is a big queue?
On Wed, 1 Mar 2000 [EMAIL PROTECTED] wrote: > Hi, I'm currently managing a mailserver for a service with around 5500 > virtual users as of this writing, and we get over 1000 new users daily. > I've been tracking our queue size since we launched, and it's currently > just shy of 1000 messages, with ~650 messages unprocessed. Is this going > to be a problem? How big are queues supposed to be? > > Also, I've lowered my queuelifetime to 2 days and increased > concurrencyremote to 120. Otherwise, it's a stock LWQ installation. Oops. Actually, it WAS a completely standard LWQ installation. I had forgotten to restart qmail-send, so it was using the concurrencyremote default of 20. I restarted qmail-send, and watched qmail-qstat report a rapidly dwindling queue. It's back to normal now. I also watched my load average shoot up to 2.5 as all the qmail-remote processes fired up. Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
How big is a big queue?
Hi, I'm currently managing a mailserver for a service with around 5500 virtual users as of this writing, and we get over 1000 new users daily. I've been tracking our queue size since we launched, and it's currently just shy of 1000 messages, with ~650 messages unprocessed. Is this going to be a problem? How big are queues supposed to be? Also, I've lowered my queuelifetime to 2 days and increased concurrencyremote to 120. Otherwise, it's a stock LWQ installation. Any helpful tips, patches, etc. would be very much appreciated. Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: Where does the bounce mail reside
On Wed, 23 Feb 2000, john wrote: > Hi, > > I would like to know where the bounced mails reside when I send it to the wrong >address in Qmail using Vpopmail. > > Could anyone let me know which directory the mail goes to when bounced. The default vpopmail installation does not bounce mail back to the sender if he tries to send to a non-existent user. Instead, the mail goes to the postmaster account. To change this, you should edit your ~vpopmail/domains//.qmail-default file, and put bounce-no-mailbox instead of the postmaster directory as a parameter to vdelivermail. HTH, Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: Custom mail directories
On Tue, 15 Feb 2000, John P. Looney wrote: > On Tue, Feb 15, 2000 at 03:21:04PM +0300, Anand Buddhdev mentioned: > > > /etc/password. I found the checkpassword program, and modified it so it > > > fills out the requesite information, and it seems to be working, for POP > > > mail. > > > However, what do I do to allow qmail to *deliver* into the custom Mail > > > directories ? > > man qmail-users > > I was hoping there was an alternate way; that I could find the home > directory required in a similar way as to how pop3d can authenticate > via a plugin. All the users info is in an SQL database, which is updated > between a hundred and a thousand times a day, so regenerating the assign > file is not an option. You probably should use something like vpopmail (http://www.inter7.com/vpopmail) which does this. The current version has support for MySQL. Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: Big and/or famous sites using qmail?
On Tue, 15 Feb 2000, S Ashok Kumar wrote: > > What is Bluetail Mail Robustifier? I couldn't find any pointers on the Net. Check out http://www.bluetail.com/products/index.shtml Very interesting reading. Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran
Re: attachment and other logging!
On Mon, 14 Feb 2000, Martin wrote: > Is it possible to figure out when a e-mail has an attachment comming in? > I have a quite slow connection, and if a person is really evil he could just send me >an attachment on 100mb. :) See the qmail-smtpd manpage and look for "databytes". Brian -- [EMAIL PROTECTED] http://www.baquiran.com AIM: bbaquiran