Paolo, In instantiated a VM to try idel=poll. Before modify kernel parameters, I would like to check current config... sadely nothing at all talk about idle: root@imsprobe0:/home/adm_astellia# cat /proc/cmdline ro root=/dev/mapper/rhel-root rd_NO_LUKS LANG=en_US.UTF-8 rd_LVM_LV=rhel/root rd_NO_MD SYSFONT=latarcyrheb-sun16 KEYBOARDTYPE=pc KEYTABLE=us crashkernel=129M@48M rhgb quiet rd_NO_DM rhgb quiet console=ttyS0 root@imsprobe0:/home/adm_astellia# sysctl -a|grep -i idle kernel.sched_domain.cpu0.domain0.idle_idx = 1 kernel.sched_domain.cpu0.domain0.newidle_idx = 0 kernel.sched_domain.cpu1.domain0.idle_idx = 1 kernel.sched_domain.cpu1.domain0.newidle_idx = 0 kernel.sched_domain.cpu2.domain0.idle_idx = 1 kernel.sched_domain.cpu2.domain0.newidle_idx = 0 kernel.sched_domain.cpu3.domain0.idle_idx = 1 kernel.sched_domain.cpu3.domain0.newidle_idx = 0 kernel.sched_domain.cpu4.domain0.idle_idx = 1 kernel.sched_domain.cpu4.domain0.newidle_idx = 0 kernel.sched_domain.cpu5.domain0.idle_idx = 1 kernel.sched_domain.cpu5.domain0.newidle_idx = 0 kernel.sched_domain.cpu6.domain0.idle_idx = 1 kernel.sched_domain.cpu6.domain0.newidle_idx = 0 net.ipv4.tcp_slow_start_after_idle = 1 Cheers, -----Message d'origine----- De : kvm-owner@xxxxxxxxxxxxxxx <kvm-owner@xxxxxxxxxxxxxxx> De la part de Gregory Esnaud Envoyé : mardi 7 janvier 2020 18:38 À : Paolo Bonzini <pbonzini@xxxxxxxxxx>; kvm@xxxxxxxxxxxxxxx Objet : RE: [Newbie question] Why usage of CPU inside VM and outside VM is different > That's a possibility. And if it's correct, your provider is certainly > correct in complaining. :) Yes, but as a cloud provider he should provide us prerequisites on VM config 😉. I will give it a try! Thanks! -----Message d'origine----- De : Paolo Bonzini <pbonzini@xxxxxxxxxx> Envoyé : mardi 7 janvier 2020 18:36 À : Gregory Esnaud <Gregory.ESNAUD@xxxxxxxx>; kvm@xxxxxxxxxxxxxxx Objet : Re: [Newbie question] Why usage of CPU inside VM and outside VM is different On 07/01/20 18:32, Gregory Esnaud wrote: > Hi Paolo, > > Thanks for your quick answer. > > Are you using only 2 CPUs at 100%, or are you using 9 CPUs at a total of 200%? > > Inside the VM, a top command show that we are using 2 CPU @ 100%: > https://framadrop.org/r/gNf3erJVf6#t2HLIpzPHGtxHzHeLrnGGjsRSZZpmv9HPVZ > t/LIIP8s= From the hypervisor (ie, *outside* the vm), a top command > show that the VM consuming 9 cpu: > https://framadrop.org/r/-FmrBhMGX5#q/6aZ8Fq2Mnuzz3+6F/pa83U+jzPox1hd98 > 4tw7DkuA= > > So, if I'm understanding you correctly, an parameter of kernel of our VM is idel=poll that we should change? > That's a possibility. And if it's correct, your provider is certainly correct in complaining. :) Paolo
