Noticing the dashboard CPU load meters are somewhat useless and not sure if this is how things are intended to be or if there is a bug somewhere.  Running a high load process (4.+ lload average) and dashboard is showing me the CPU is idle/low load and not much activity.  This is misleading and can be confusing if I didn't already know several processes were running that had high CPU loading.  Suggest/request Linux 1, 5, 15 minute load average display be shown and perhaps %wa and any other useful data available to decipher between CPU load and I/O load.

 

 

 

 

 

Edited May 20, 20179 yr by unevent

To put the CPU cores to a high load use the following command over a ssh-session:

cat /dev/urandom > /dev/null

Open one ssh-session per core (physical and virtual) and see, what happens.

Edited May 20, 20179 yr by Zonediver

Although I understand where you are coming from, load averages are far more confusing, because a load average of say 4 on your system means (treating hyperthreaded cores as real cores) means 50% CPU utilization, on my system the same load average means 100%, and for many users here would mean 13%...  IE: The same load average means something different depending upon which CPU you have installed.  Not to mention that joe average user understands that 100% = 100% but would actually need to google (and understand) and perform the math to determine what the load average actually means to them.

 

Mind you, FCP does for instance in its troubleshooting mode report the load average as its very useful when trying to determine out of control processes causing issues.

Edited May 20, 20179 yr by Squid

6 minutes ago, Squid said:

Although I understand where you are coming from, load averages are far more confusing, because a load average of say 4 on your system means (treating hyperthreaded cores as real cores) means 50% CPU utilization, on my system the same load average means 100%, and for many users here would mean 13%...  IE: The same load average means something different depending upon which CPU you have installed.  Not to mention that joe average user understands that 100% = 100% but would actually need to google (and understand) and perform the math to determine what the load average actually means to them.

 

Mind you, FCP does for instance in its troubleshooting mode report the load average as its very useful when trying to determine out of control processes causing issues.

 

Completely agree it would be confusing for those that don't understand what the load average means and how to interpret it based on CPU type.  Also by itself it cannot differentiate CPU load vs I/O load.  I suppose perhaps an advanced tab on the stats page where you have standard 'top' info along with any other useful data.

Just now, unevent said:

 

Completely agree it would be confusing for those that don't understand what the load average means and how to interpret it based on CPU type.  Also by itself it cannot differentiate CPU load vs I/O load.  I suppose perhaps an advanced tab on the stats page where you have standard 'top' info along with any other useful data.

That I'll agree with...