I need some expert thoughts about my sendmail config. ...

[someone_4]

I have a mail server L-200 with two 550mhz cpu , for a mid sized isp with 15,204 mail boxes. All in /var/mail filesystem. We had bad disk bottlenecks so I was able to implement a separate file system for the mailq /var/mail/mqueue and span the /var/mail filesystem over two disk. Which helped out a little. I am using sendmail 8.12.10 and a qpopper 2.53. This was all set up before I got here involved with the configuration and at the time the company was looking to do this cheap and fast.

The issue I am having now is that I have a cpu and disk bottle neck again. The system is so slow and people are starting to complain about connection issues. I/O is high and the to shows idle 10 - 25%. Even doing regular commands like cd and ls are slow with allot of lag. I feel that the server is just bogged down and cant handle the amount of traffic that it is getting. I have been pushing a new mailserver solution running Communigate software for about a year. And none of the top management has taken me serious when I told them we would start having mail issues. But the issue now is that everyone is beating me up wanting me to fix this issue now. And everyone wants to know why this is happening. And what I can do to fix it. I really feel like my hands are tied. If you as me the server was just not set up properly for an enterprise level solution.

At this point what I am looking for is the opinion of somoene who knows sendmail systems up and down. And hopefully can give me something to bring to the table as far as why we would have these issues and why we should move to a new configuration.

Thanks
Richard

[someone_4]

on another note.
I feel that the issue is too many users and too much traffic for the server to handle with just a plain sendmail solution. Over 50% is spam is it is just eatting up resources.

I have tried to tell them that the disk has high I/O and that the cpu is high. But they want to know why!! And what I can do to fix it now! And want a more specific answer then just because there are too many users. And that there is realy nothing I can do to fix it.

Richard

[James Specht · Accepted solution]

I hate to say it this way but it's not sendmail's fault. It's the hardware that is your problem. Sendmail, or any mail server is going to require disk I/O and cpu to do its job. Sendmail is pretty effective in my humble opinion. I currently run sendmail as our central mail hub for about 70,000 users. All outgoing and incoming mail passes through to the tune of 2 - 5 gigs of traffic a week.

In your case though you are housing the mail and this adds to your I/O traffic. My suggestion for the quick fix, if possible, would be to buy some disk and stripe your /var. If possible use dual controllers to the different disks to help open up the bottle neck.

I'm not sure if the L-200's 550 CPUs are the max either. Check the specs and if so there might be some used CPUs you could upgrade to for a reasonable price.

Good Luck!

--Jim