I don't think these issue have much to do with Spice, but with the
amount of memory oVirt sets to VMs by default, which in some cases for
desktop usage seems too little. A field where that could be adjusted
without having to edit files in the Engine would probably resolve this
issue, or am I missing anything ?
On 07/03/2018 15:43, Michal Skrivanek wrote:
On 7 Mar 2018, at 14:03, FERNANDO FREDIANI <fernando.fredi...@upx.com
Resurrecting this topic. I made the changes as per your instructions
below on the Engine configuration but it had no effect on the VM
graphics memory. Is it necessary to restart the Engine after adding
the 20-overload.properties file ? Also I don't think is necessary to
do any changes on the hosts right ?
correct on both
there were no changes recently, these are the most recent guidelines
we got from SPICE people. They might be out of date. Would be good to
raise that specifically (the performance difference for default sizes)
to them, can you narrow it down and post to
On the recent updates has anything changed in the terms on how to
change the video memory assigned to any given VM. I guess it is
something that has been forgotten overtime, specially if you are
running a VDI-like environment whcih depends very much on the video
Let me know.
On 24/11/2017 20:45, Gianluca Cecchi wrote:
On Fri, Nov 24, 2017 at 5:50 PM, FERNANDO FREDIANI
<fernando.fredi...@upx.com <mailto:fernando.fredi...@upx.com>> wrote:
I have made a Export of the same VM created in oVirt to a server
running pure qemu/KVM and which creates new VMs profiles with
vram 65536 and it turned on the Windows 10 which run perfectly
with that configuration.
Was reading some documentation that it may be possible to change
the file /usr/share/ovirt-engine/conf/osinfo-defaults.properties
in order to change it for the profile you want but I am not sure
how these changed should be made if directly in that file, on
another one just with custom configs and also how to apply them
immediatelly to any new or existing VM ? I am pretty confident
once vram is increased that should resolve the issue with not
only Windows 10 VMs, but other as well.
Anyone can give a hint about the correct procedure to apply this
Thanks in advance.
based on this:
you should create a file of kind
but I think you can only overwrite the multiplier and not directly
the vgamem (or vgamem_mb in rhel 7) values
so that you could put something like this inside it:
os.windows_10.devices.display.vramMultiplier.value = 2
os.windows_10x64.devices.display.vramMultiplier.value = 2
I think there are no values for vgamem_mb
I found these two threads in 2016
that confirms you cannot set vgamem
that suggests to create a hook
Just a hack that came into mind:
in a CentOS vm of mine in a 4.1.5 environment I see that by default
I get this qemu command line
Based on this:
vgamem = 16 MB * number_of_heads
I verified that if I edit the vm in the gui and set Monitors=4 in
console section (but with the aim of using only the first head) and
then I power off and power on the VM, I get now
I have not a client to connect and verify any improvement: I don't
know if you will be able to use all the new ram in the only first
head with a better experience or if it is partitioned in some way...
Could you try eventually?
Users mailing list
Users mailing list