I set the Kernel parameters, put the host in maintenance, rebooted it, then
did a "reinstall host", but its still showing any available in the host
general info.

[image: image.png]

[image: image.png]

Running Node 4.2.8,  should the kernel parameter be enough, or do I need to
edit a conf file on the host?


On Fri, Feb 15, 2019 at 9:41 AM Michal Skrivanek <
[email protected]> wrote:

>
>
> On 15 Feb 2019, at 18:17, Vincent Royer <[email protected]> wrote:
>
> Thanks, I saw that article and it is a hit helpful, but still I don't see
> how I can tell how many huge pages and of which size my host supports in
> the first place.
>
>
> 2M and 1G on x86
>
>
> Do all hosts support a certain number, is this calculated by available
> RAM?
>
>
> No, it’s just what linux kernel and hw architecture support
>
>
> How do I know what size and amount to specify in the kernel command line?
>
>
> Amount is on you. Hugepages are generally unusable for other normal apps.
> It’s a matter of flexibility. You need to leave enough available 4k pages
> for maintenance processes for OS to work. I’d leave few GBs at least...
> Or if you plan to use it for just one VM and not for others then just size
> it according to that VM’s needs
>
> Thanks,
> michal
>
>
>
>
> On Fri, Feb 15, 2019, 2:35 AM Martin Sivak <[email protected] wrote:
>
>> Hi,
>>
>> the hugepages configuration has two aspects - hugepage size and count.
>> You can have multiple sizes at the same time. I am not sure what the
>> recommended configuration is or how to pass it to kernel so others will
>> have to help here (I think 2M and 1G are the usual sizes).
>>
>> The VM memory configuration is then taken from two places. The custom
>> property tells oVirt which size of hugepages you want to use (in KiB iirc)
>> and the actual number of used hugepages is computed from the total memory
>> configured for the VM at the usual place (it has to be a multiple of the
>> selected hugepage size).
>>
>> One of our colleagues wrote a blog post about this:
>> https://mpolednik.github.io/2017/06/26/hugepages-and-ovirt/
>>
>> Best regards
>>
>> Martin Sivak
>>
>> On Fri, Feb 15, 2019 at 5:35 AM Vincent Royer <[email protected]>
>> wrote:
>>
>>> How do I know how many huge pages my hosts can support?
>>>
>>> cat /proc/meminfo | grep Huge
>>>  AnonHugePages:  17684480 kB
>>> HugePages_Total:       0
>>> HugePages_Free:        0
>>> HugePages_Rsvd:        0
>>> HugePages_Surp:        0
>>> Hugepagesize:       2048 kB
>>>
>>> [image: image.png]
>>>
>>> And once I know, I set the kernel parameters here, and reboot the host,
>>> correct?
>>>
>>> [image: image.png]
>>>
>>>
>>> And then I assume I assign them to the VM here?  How do I decide how
>>> many huge pages and what size a particular VM can benefit from?
>>>
>>> <image.png>
>>>
>>>
>>> Is there a part of the docs I am not finding that covers this?
>>> _______________________________________________
>>> Users mailing list -- [email protected]
>>> To unsubscribe send an email to [email protected]
>>> 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/[email protected]/message/ZMGN7YZXSQ5UUZ43U5RPSXTZOTYI2XAZ/
>>>
>> _______________________________________________
> Users mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
> 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/[email protected]/message/O43AJBFDCCDCDGB53IZRH5ROFX4FWPIF/
>
>
_______________________________________________
Users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
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/[email protected]/message/HTQEMJU3QYL77H256MRZNO75ZD3CHUWH/

Reply via email to