I have a dedicated server, I try my best to have always it up and running. But last week, it's been down for 2 hours, 1.5 of which I had no idea because basically, no customer had posted a ticket. When I got the first phone call, basically it had been down for 1h30. It took almost half an hour to find out who on the server was doing the mess and to stop the offense. CPU and RAM usage was full.

How to avoid such problems? Or how to minimize them?
Thanks.

you need some type of monitoring..something like Servers Alive or alertbot.com

I use site uptime 5m monitoring. Alert goes to my cell when server is down.

You could also try to program some sort of script to detect failure and restart the web server.

There are a couple things you can do to guarantee you better uptime.

1) as FluffyTigger mentioned, SiteUptime.com (http://www.siteuptime.com/) monitors every 30 minutes for free, or 5Minutes for $5/month. This will allert an email address. I would suggest forwarding this to your cell phone in the form of a text message for best response.

2) Double your server count. Set up properly, you will have a backup server that will kick in as soon as something happens to your primary server. A cheaper alternative to this is a Double hard drive system set up in RAID configuration.

Unfortunately, with 100% uptime comes $. However the more measures you take, the less you'll find downtime is an issue.

Thanks. I already have Raid-1 set up here ;)

Site Alive doesn't send any alert even if the site is down. It says:

Up: 0
Down: 1

But it doesn't send any alert. I set 3 types of alert: Bell, Mail and IM.

Places like SeeksAdmin will know your server is down within 5minutes or less and will bring it back up.

Use Hyperspin.com - they will send an email (which you can setup to go to your cell phone to page you).

--Tina

You could have someone keep an eye on your server like everyone else said or you could have someone help take care of your server so if it goes offline he can contact you or try to bring it back online themselves.

Use Hyperspin.com - they will send an email (which you can setup to go to your cell phone to page you).

Thanks for this link Tina. Just spent awhile exploring their site, and they look like they really have their act together, was impressed. Appreciate it.

You could setup a monitoring service yourself using nagios.org.

Plus that would allow you to set which action should be taken to rectify if a problem occured, so f.e nagios can restart a server/process automatically and send you an e-mail with the result, ruther than bothering you every 2 minutes that the service is down.

It can also "analyze" what the actually issue is and point you directly to the issue instead of just saying "your server is unreachable again" it would say that the network connection to your default gateway is broken.

That might be really what you are after. The uptime reports will be created of course in the background.

I think you might find Nagios worth to look at.

Use Hyperspin.com - they will send an email (which you can setup to go to your cell phone to page you).

--Tina

Agreed - hyperspin is great, ESPECIALLY their cell phone page feature

2) Double your server count. Set up properly, you will have a backup server that will kick in as soon as something happens to your primary server. A cheaper alternative to this is a Double hard drive system set up in RAID configuration.

RAID will not keep the sites online UNLESS it is a harddrive problem, which 95% of the time it is not.

In which case, 5% of the time its hardware it WILL help.

I use alertbot it works pretty good.