Skip to content
Permo Docs

Server load average

The load average shows how much work the server had over the last few minutes: how many tasks were running on the processors or waiting for them, on average. Linux reports three values, averaged over 1, 5 and 15 minutes. It also counts tasks that are waiting for the disk.

Why it matters

Section titled “Why it matters”

When more tasks want to run than there are processors, they queue, and every page that WordPress builds takes longer. The load average shows this pressure before visitors notice it.

The number only means something next to the number of CPU cores. A load of 4 keeps a server with four cores fully busy; on a server with two, half the tasks are waiting. That’s why Permo also calculates load per CPU core. The load average isn’t a percentage, and it doesn’t say which site or program caused the work.

The three values together show the direction: a high 1-minute value with a low 15-minute value is a short peak; three high values mean the server has been busy for a while.

What influences it

Section titled “What influences it”

WordPress: The most common cause of a high load on a WordPress server is many PHP requests at the same time that the page cache doesn’t answer: crawlers and bots going through the site, logged-in users, the cart and checkout of a shop, the admin area. Scheduled tasks add bursts of work: backups, imports, image processing, newsletter sending.

Server: Everything else running on the same machine counts too: the database, other sites on a shared server, the provider’s own tools. Slow storage raises the load because tasks waiting for the disk are included. On some virtual servers and containers, the value reflects the whole host rather than your share of it.

Cache: Every request a page cache answers is a request PHP doesn’t have to handle, so a working page cache keeps the load down.

In Permo

Section titled “In Permo”

The Load average card is in System info, tab Server & System, under Server. It shows the three values with the note 1, 5 and 15 minutes, as of Permo’s last check of the server; the tab says when that was. The card gives context; Permo doesn’t rate it. The rating is on the Server load card, as load per CPU core.

Permo also reads the three values at the start of every check; the History, view Server, shows the 1-minute value per CPU core over time. On Windows servers, which don’t provide a load average, the values stay empty and the card says This server doesn’t report its load.

Section titled “Related”

Sources