Solarbird 

this was what our web server was looking like a couple of weeks ago when i went into emergency perf optimisation/reconfiguration mode and also ordered a new-to-us used CPU with two more cores and more than double the threads that still fit in our motherboard's socket after a post on my federated blog went low-key viral (55,000ish "impressions")

i like the load average of 60.10 but i particularly like the mariadb task using 1182% of CPU

i don't actually know what that means and i think top was confused

#KRANG #murknet #ohno #cpu #apache #mariadb

Jun 7 at 4:55amWeb
Show threadIngvarJun 7
@moira I read 1182% CPU as "11.82 core-seconds / second". But, not knowing how many cores (or threads, I guess) that machine has, I have no idea if that is plausible.
Show threadTroubleJun 7
@vatine @moira You are exactly right! You can exceed 100% with a multi-threaded program that's one process, using multiple cores simultaneously.
Note that this is impossible to do in Python due to the GIL, despite the terminology; only one thread is allowed to execute at a time.
Show threadSolarbird Jun 7

@trouble @vatine but that old CPU only had six cores (and threads) so

confusion

Show threadIngvarJun 7
@moira @trouble Then I have no plausible explanation, other than "top, or the kernel, seems confused".
Show threadSolarbird Jun 7

@vatine @trouble top

and/or the kernel

and me. definitely me

Show threadMarcos DioneJun 7
@moira load represents a mix of many reasons why your system is sluggish. You may be better looking at pressure values to know which combination of RAM, CPU and disk are you lagging.
Show threadSolarbird Jun 7
@mdione yeah i know the point was this was _before_ i did all the perf/optimisation work