high load on hosts

I'm wondering how I can best limit the ability of VMs to overrun the load on hosts. I have a fairly stock 4.2 HCI setup with three well spec'ed servers, 10Gbe/SSDs, plenty of ram and CPU with only a hand full of light use VMs. I notice when the occasional demanding job is run on a VM I'm seeing load average on host node shoot up in to the 20-30s, how can a single "medium" vm cause host load to rise so high?

I am not trying to be sarcastic here, but the host resources are controlled by what you allocate to the vm... that is kinda how virtualization works On Fri, Oct 19, 2018 at 8:47 AM Jayme <jaymef@gmail.com> wrote:
I'm wondering how I can best limit the ability of VMs to overrun the load on hosts. I have a fairly stock 4.2 HCI setup with three well spec'ed servers, 10Gbe/SSDs, plenty of ram and CPU with only a hand full of light use VMs. I notice when the occasional demanding job is run on a VM I'm seeing load average on host node shoot up in to the 20-30s, how can a single "medium" vm cause host load to rise so high?
_______________________________________________ Users mailing list -- users@ovirt.org To unsubscribe send an email to users-leave@ovirt.org Privacy Statement: https://www.ovirt.org/site/privacy-policy/ oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/ List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/LVSVFSI2KXURMX...

Also the nature of Converged compute and storage will cause more usage on a host than you are used to seeing... Storage costs cpu/ram cycles too On Sat, Oct 20, 2018 at 7:29 PM Donny Davis <donny@fortnebula.com> wrote:
I am not trying to be sarcastic here, but the host resources are controlled by what you allocate to the vm... that is kinda how virtualization works
On Fri, Oct 19, 2018 at 8:47 AM Jayme <jaymef@gmail.com> wrote:
I'm wondering how I can best limit the ability of VMs to overrun the load on hosts. I have a fairly stock 4.2 HCI setup with three well spec'ed servers, 10Gbe/SSDs, plenty of ram and CPU with only a hand full of light use VMs. I notice when the occasional demanding job is run on a VM I'm seeing load average on host node shoot up in to the 20-30s, how can a single "medium" vm cause host load to rise so high?
_______________________________________________ Users mailing list -- users@ovirt.org To unsubscribe send an email to users-leave@ovirt.org Privacy Statement: https://www.ovirt.org/site/privacy-policy/ oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/ List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/LVSVFSI2KXURMX...

Yes I understand how virtualization works, but what I'm saying is that my hosts each have 16 cores, 256gb ram, 10GBe networking and all SSD storage with VMs allocation only a fraction of the available resources, yet high load on a single VM with 2 cpus and 4gb ram allocated can cause the host load to rise above 30. I'm trying to determine why one VM with limited resources can cause the load on the host to go up so much. On Sat, Oct 20, 2018 at 8:32 PM Donny Davis <donny@fortnebula.com> wrote:
Also the nature of Converged compute and storage will cause more usage on a host than you are used to seeing... Storage costs cpu/ram cycles too
On Sat, Oct 20, 2018 at 7:29 PM Donny Davis <donny@fortnebula.com> wrote:
I am not trying to be sarcastic here, but the host resources are controlled by what you allocate to the vm... that is kinda how virtualization works
On Fri, Oct 19, 2018 at 8:47 AM Jayme <jaymef@gmail.com> wrote:
I'm wondering how I can best limit the ability of VMs to overrun the load on hosts. I have a fairly stock 4.2 HCI setup with three well spec'ed servers, 10Gbe/SSDs, plenty of ram and CPU with only a hand full of light use VMs. I notice when the occasional demanding job is run on a VM I'm seeing load average on host node shoot up in to the 20-30s, how can a single "medium" vm cause host load to rise so high?
_______________________________________________ Users mailing list -- users@ovirt.org To unsubscribe send an email to users-leave@ovirt.org Privacy Statement: https://www.ovirt.org/site/privacy-policy/ oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/ List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/LVSVFSI2KXURMX...

What is the workload from the VM? Is it an IO bound workload? On Sun, Oct 21, 2018 at 9:19 AM Jayme <jaymef@gmail.com> wrote:
Yes I understand how virtualization works, but what I'm saying is that my hosts each have 16 cores, 256gb ram, 10GBe networking and all SSD storage with VMs allocation only a fraction of the available resources, yet high load on a single VM with 2 cpus and 4gb ram allocated can cause the host load to rise above 30. I'm trying to determine why one VM with limited resources can cause the load on the host to go up so much.
On Sat, Oct 20, 2018 at 8:32 PM Donny Davis <donny@fortnebula.com> wrote:
Also the nature of Converged compute and storage will cause more usage on a host than you are used to seeing... Storage costs cpu/ram cycles too
On Sat, Oct 20, 2018 at 7:29 PM Donny Davis <donny@fortnebula.com> wrote:
I am not trying to be sarcastic here, but the host resources are controlled by what you allocate to the vm... that is kinda how virtualization works
On Fri, Oct 19, 2018 at 8:47 AM Jayme <jaymef@gmail.com> wrote:
I'm wondering how I can best limit the ability of VMs to overrun the load on hosts. I have a fairly stock 4.2 HCI setup with three well spec'ed servers, 10Gbe/SSDs, plenty of ram and CPU with only a hand full of light use VMs. I notice when the occasional demanding job is run on a VM I'm seeing load average on host node shoot up in to the 20-30s, how can a single "medium" vm cause host load to rise so high?
_______________________________________________ Users mailing list -- users@ovirt.org To unsubscribe send an email to users-leave@ovirt.org Privacy Statement: https://www.ovirt.org/site/privacy-policy/ oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/ List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/LVSVFSI2KXURMX...

Yes, you are right, load 30 with a 2 CPU VM is nearly impossible. In the past, there where a issue with a defect HD and a failing resync on glusterfs here in the list. - Take a look deeper look on your ovirt-node(s) -- Systemcalls with (seeing kernel und userspace syscalls): # perf top - IO with: # iotop - Processes with: # htop Ansible Script to install the tools: --- # # Install Additional Packages on Centos 7 machines # Possible Repos: base,updates,extras,centosplus # - hosts: - ovirt-nodes gather_facts: False tasks: - name: Install additional Packages yum: enablerepo: base name: - perf - iotop state: present - name: Install additional remote Packages yum: name: - https://kojipkgs.fedoraproject.org/packages/htop/2.2.0/1.el7/x86_64/htop-2.2... state: present ... On 10/19/18 2:46 PM, Jayme wrote:
I'm wondering how I can best limit the ability of VMs to overrun the load on hosts. I have a fairly stock 4.2 HCI setup with three well spec'ed servers, 10Gbe/SSDs, plenty of ram and CPU with only a hand full of light use VMs. I notice when the occasional demanding job is run on a VM I'm seeing load average on host node shoot up in to the 20-30s, how can a single "medium" vm cause host load to rise so high?
_______________________________________________ Users mailing list -- users@ovirt.org To unsubscribe send an email to users-leave@ovirt.org Privacy Statement: https://www.ovirt.org/site/privacy-policy/ oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/ List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/LVSVFSI2KXURMX...

Le 23 oct. 2018 à 16:30, Oliver Riesener <Oliver.Riesener@hs-bremen.de> a écrit :
Yes, you are right, load 30 with a 2 CPU VM is nearly impossible.
Load on linux is a poor indicator, Brendan explan why: http://www.brendangregg.com/blog/2017-08-08/linux-load-averages.html
participants (4)
-
Donny Davis
-
Fabrice Bacchella
-
Jayme
-
Oliver Riesener