Although the two servers are fairly similar, my loads seem to be a fair bit higher on my Virtualmin server than on my cPanel server (same sites moved across) and so wondered what you guys think.
Both running near enough the same versions of PHP and mysql. Loads rarely went over 0.20 on my old server, sometimes going to 0.40, but on my new server they seem to be around 0.40 and can peak to 0.80 or even above 1.
I’m just wondering if there’s anything I can do to help it? I’ve already deleted Awstats and Webalizer from the webmin modules, is there anything else that could probably go?
I’ve also switched to mod_php as I hear it’s meant to be faster (all the sites on the server are my own - so don’t really need the security of fcgiphp).
One of the main reasons I switched to Virtualmin was because I heard it wasn’t a resource hog like cPanel, so must admit I was hoping to see the loads lower even tho the server has a slower CPU.
Well, Virtualmin and cPanel aren’t likely to be what’s causing the load on either of your systems, it’s the other services you’re running… Apache, PHP, Email, and so forth.
So the key would be to review what’s using resources on your system, and then from there to determine how to reduce the load.
Also, the Linux distribution you’re using is capable of making a difference – which Linux distribution/version are you running on both of those systems?
Now, I’ll offer that a load of 1.0 and below is pretty decent – and I’m not sure that I’d be concerned at all about a server with loads of 0.20 or 0.40.
There’s nothing wrong with optimizing though. You’re right that mod_php is fast, that may be a good option to use in your case.
What you may want to do is run the command “top” – and monitor which processes are the heavy resource users.
You also may want to disable any services you aren’t using. If you don’t need Mailman, for example, you’d want to stop the mailman process. And the same for any other services you don’t need.
You can see what services are being started at boot time in Webmin -> Bootup and Shutdown.
Virtualmin has more features enabled, by default, particularly in the mail stack. Spam and virus filtering are on, by default, in a Virtualmin system which I don’t think is true in a cPanel system (I’m working on a migration that has 32GB of mailboxes that are all spam, so I’m assuming spam filtering is either not enabled by default, or it sucks by default, on a cPanel system). Depending on how those services are being run (daemonized leads to much lower load, but slightly more memory usage), it could account for the difference you see in load.
But, as Eric notes, the control panel itself is irrelevant WRT your system load. All the other components are similar between cPanel and Virtualmin, so it’s going to be configuration differences, and possibly additional features in the Virtualmin system, that make the difference.
I’ve been watching ‘top’ for the last few days, and I think the httpd and mysqld process run about the same as they did on the old server (all less than 10% cpu). Both servers are running CentOS 5.8.
The two things that stand out on the Virtualmin server are the collectinfo.pl process (using up to 50% cpu) and the miniserv processes (sometimes 15%, but usually a lot less). I have changed the cron settings for collectinfo.pl to run 3 times a day instead of every 10 or so minutes that it was set to run. Not sure what to do about miniserv - if I switch it off will that mean I can’t log-in to Webmin?
I have switched off mailman too - can you think of anything else that I might not need?
I looked in the Bootup and Shutdown page and these don’t seem necessary but I am not entirely sure:
Usermin - I don’t use webmail, the only web interface I use is the Webmin/Virtualmin/:10000 log-in, so can I safely switch this off?
hidd - (Bluetooth Human Interface Device Daemon) I highly doubt the DC will need this, so can I turn it off?
lookup-domain (Daemon for quickly looking up Virtualmin servers from procmail) Do I need this? I’m using postfix and dovcot for email.
proftpd - I don’t use FTP, just SFTP, so can I switch proftpd off?
Anything else you might guess that I might not need? If you can think of any I can verify when I think I might need them or not.
Thanks Eric!
Edit: I forgot to say, the logwatch that I get emailed to me is huge compared to the one I get from my cPanel server - does all that is being logged really need to be logged? Such as returned emails, failed pop logins etc.
Edit2: I have also turned off monitor.pl as I don’t use ‘status monitoring’ per virtual server (after reading this thread: http://www.virtualmin.com/node/17804)
You can disable Usermin if you don’t need webmail, and ProFTPd if you aren’t using FTP. I don’t know much about hidd, you could experiment with disabling if and see if everything works afterwards
I believe you can disable lookup-domain as well – although that’s needed for email delivery, it’ll fall back to using a process that runs each time an email is received, rather than a daemon.
the logwatch that I get emailed to me is huge compared to the one I get from my cPanel server - does all that is being logged really need to be logged? Such as returned emails, failed pop logins etc.
Well, the whole idea of logging is to log when events happen
If you don’t want to know about those things, you can always disable logwatch, or tune it to be more what you want – but I don’t recommend turning off failed events like you’re seeing, as those become really useful when it comes time to troubleshoot problems.
I can tel you that as far as I know miniserv is the server that webmin uses internally, so I do believe that yes-you will be stuck if you power that down.
Turning off ProFTPd will be safe to do if you aren’t using it.
If you are not going to be using usermin, or, if you have some other webmail client like Squirrelmail in mind, feel free to power that off, I’m sure that it will free up a little bit of resources.
If you don’t plan on other services, try turning them off, and testing to make sure things still function as you expect.
If you mess something up, see if turning on whatever you just shut off that caused the problem fixes things.
Letting us know what OS you are using as Eric asked for earlier in this posting would be a useful thing.
CentOS, Debian5/6, Ubuntu10.04LTS or other will be useful as well.
Hope this helps.
Well, I ended up moving back to my old server, but got the hd replaced with two 500GBs and is now in Raid 1 as the new server. The main reason for moving back was because hotmail had black-listed the IPs on the new server, tried to get them de-listed but to no avail. So thought I’d move back to the old server with the IPs I know are safe.
Good news is the loads are MUCH better back on the old server, although I haven’t added the my.cnf tweaks yet. The only thing I’ve done so far is change collectinfo.pl to run just once a day - do you think that’s ok? Probably didn’t need to now.
Yup, it’s perfectly fine to have collectinfo.pl running once a day, that just updates various statistics. You can also set bw.pl (the bandwidth monitoring) to run less frequently as well.
If you don’t see bw.pl in there, that’s okay, it means it’s not active, which is even better for performance
You can disable the cron jobs for anything you don’t need. That likely includes that awstats cron job you mentioned. I don’t know what mcelog.cron is, that’s not something that’s installed by default.
