If you’ve ever looked at htop, you probably wondered what the load average is.
At its most basic level, what it shows is how stressed out your CPU is. The ideal load average is less than 1 per core. So if you have a quad core system, a load average of under 4 means your system has more than enough CPU power for your task. If it’s exactly the number of cores you have, it’s reaching its limits, but can still deal with it. If it goes beyond one per core, for only the shortest measurement, more on that in a bit, it’s fine. If all three load averages are above the number of cores you have.
Why are there three numbers?
If you look in htop, or uptime, there are three different load averages. These are the average load over different period of times. The first one taken over the least time, the middle one is taken over more time, etc. The exact numbers are 1 minute for the first one, 5 minutes for the middle one, and 15 minutes for the last one. So if it’s above the core number for just a minute, you’ll be fine, it’s probably just a spike in traffic.
Why is this important?
The load average is a good measurement of how stressed your server is. If it constantly is above the recommended value, then you should upgrade your server. If you use a VPS, some providers like Vultr, allow you to upgrade your VPS in the settings. If you can’t upgrade your server, then use a tool like htop to see what is using the most resources, and how to reduce its resource usage. Also, processes in Disk Sleep will make your load average increase by a lot, so that’s also something to check for.
BLOG UPDATE: I have now switched to posting every other day in order to focus more time on my other blog(s).