From owner-freebsd-isp Wed Jun 4 13:41:25 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id NAA07735 for isp-outgoing; Wed, 4 Jun 1997 13:41:25 -0700 (PDT) Received: from duke.neuronet.com.my (neuronet.com.my [202.184.153.3]) by hub.freebsd.org (8.8.5/8.8.5) with SMTP id NAA07729 for ; Wed, 4 Jun 1997 13:41:21 -0700 (PDT) From: sweeting@neuronet.com.my Received: from love.com.my by duke.neuronet.com.my; (5.65v3.2/1.1.8.2/25Jul96-0519PM) id AA18501; Thu, 5 Jun 1997 04:39:24 +0800 Date: Thu, 5 Jun 1997 04:39:24 +0800 Message-Id: <3.0.32.19970605043257.00dec3a0@neuronet.com.my> X-Sender: sweeting@neuronet.com.my X-Mailer: Windows Eudora Pro Version 3.0 (32) To: freebsd-isp@freebsd.org Subject: help pls : how to optimise mail server response Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: owner-isp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I apologise for asking this question again but had little response last time and now it has become a serious problem since a client is complaining as a result of the following : The problem is that they are having a lot of difficulty collecting mail from the mailhub I set up for them on a P166 FreeBSD box with 64 MB RAM. We are a webcompany with 2 MB line to the same ISP that our customer is dialling into. (We are not the ISP) Een over a LAN, when I ping the FreeBSD box, the very first request _always_ times out. (Almost as if the machine is waking up) So, the only causes of trouble I can think of are : 1) network card. (it's a 3Com). which do you all recommend ? 2) DNS - is handled by a DEC Alpha 500 (runnign DU) which is possibly in need of RAM (it's handling mail, and 4 webservers, X term, and DNS for 50 domains.... somebody once told me that each domain should have 1 MB RAM.) 3) Jammed network at the ISP 4) I configured it inefficiently (config files below) In order to reduce the timeouts and inability to contact the mailserver (2 complaints I am receiving), what action would you take ? I'm already considering : 1) changing the network card 2) move the DNS to the same machine 3) increasing the timeouts .... but I think that that is not really a satisfactory solution but rather avoiding the real problem. Apologies since this must seem so pathetic compared to the huge systems most of you are running without problem. Thank you very much in advance, Chas ---------------------------- configuration : A DEC box is running the webserver and virtual domain for our client : "company.com.my" xxx.xxx.xxx.aaa The FreeBSD box "mailhub.com.my" xxx.xxx.xxx.bbb receives mail for "mail.company.com.my" by using the changes to rule S98 in sendmail. == Zone file for "company.com.my" == company.com.my. IN SOA us.com.my. postmaster.us.com.my. ( 23 ; Serial 300 ; Refresh - 5 minutes 60 ; Retry - 1 minute 1209600 ; Expire - 2 weeks 43200 ) ; Minimum - 12 hours ; IN NS us.com.my. IN NS relay1.jaring.my. IN MX 10 mail IN A xxx.xxx.xxx.aaa www IN CNAME company.com.my. mail IN MX 10 mail mail IN A xxx.xxx.xxx.bbb == user mapping in sendmail.db on the freebsd mail hub == user1@company.com.my user1 user2@company.com.my user2 user3@company.com.my user3 user4@company.com.my user4 ..etc == clipped from sendmail.cf == Cwlocalhost # file containing names of hosts for which we receive email Fw-o /etc/sendmail.cw == clipped from sendmail.cw == company.com.my mail.company.com.my