Unsuspecting people:Me: *I wonder if I can have a causal conversation about something that's been bugging me lately about CFS in the #Linux kernel with anyone here.*
-
Scott Williams 🐧wrote on last edited by [email protected]
Unsuspecting people:
Me: *I wonder if I can have a casual conversation about something that's been bugging me lately about CFS in the #Linux kernel with anyone here.*
-
Scott Williams 🐧replied to Scott Williams 🐧 on last edited by
"I'd like your thoughts on if CFS can be completely fair if its behavior depends so much on the version and distribution of the kernel. Let's compare Google's current cOS vs SLES 15SP5, for example."
-
@[email protected] said in Unsuspecting people:Me: *I wonder if I can have a causal conversation about something that's been bugging me lately about CFS in the #Linux kernel with anyone here.*:
I'd like your thoughts on if CFS can be completely fair if its behavior depends so much on the version and distribution of the kernel. Let's compare Google's current cOS vs SLES 15SP5, for example.
cOS prioritizes containerized workloads, ensuring they get more CPU time compared to other processes running on the system. This aligns with Google's focus on containerized applications in cloud environments.
So the rulebook of whats most fair is almost like a George Orwell quote,
all processes are equal, but some are more equal than others! -
@eeeee I was very surprised to use the same config in GKE elsewhere to find that cOS is far more lenient about throttling to cpu limits than most every other kernel.
