Probably best to not change those settings when the machine is running. When it is not running, click on “View” and select “Details”. It lists the hardware and shows what you can change.
But, can virtual guest clock be changed to UTC or local?
Specifically when talking about configuring using libvirt (may be different in other situations),
Memory “current” and “maximum” settings do refer to the memory range for ballooning. But, you should know that “current” is only an approximate setting and in various states the virtual machine may or is likely using a different amount of memory than whatever is the “current” setting. This generally requires virtio, but as stated in a previous post in this thread, whatever driver you need should automatically be used.
NTP is only a protocol, and setting the clock to use NTP settings only means that you are advised to configure remote NTP servers for your clock reference and not to reference the HostOS or BIOS clock. This has nothing to do with setting the time zone offset, which you would do separately. This is particularly important if you do anything that requires accurate time, like implementing network security(ie Multiple VMs and/or HostOS in a same Workgroup or Domain).
So “current” memory setting is an approximated value, i.e., if I set it to 2 Gb it could actually go to 2136, 1967 Mb… or so. But my question is if these libvirt’s “current” and “maximum” memory settings really do automatic or “dynamic” memory ballooning.
I’m following these “best” advises: https://pve.proxmox.com/wiki/Windows_10_guest_best_practices
I know it’s a different distribution, but from what I’ve read it’s kind of an “authority” in virtualization world, so well used as a good/due reference for everything.
Also, how can one change VM’s RTC offset to local or UTC from libvirt?
Current doesn’t even mean “approximae,” but it might mean something like “maybe something close.”
Practically speaking, you can more or less think of it as “something much lower than the maximum.”
Regardless whatever the setting, the system will allocate something adequate when your VM starts no matter what it might be, and at various times like if your system is restarting from suspend. The point is that the system will not depend on your setting, but will manage your machine according to what it needs and wants.
In any case, it’s inadvisable to rely on memory ballooning, it should be available or configured only for exceptional circumstances. The SOP should be to know what your machine may require and fix both “current” and “maximum” to the that amount (a single, same number).
In the special case of Windows Guests, virtio drivers must be installed for certain “advanced” features. Install and configure them as described in your reference.
You don’t configure your VM’s time in the VM settings, it’s the ordinary setup in the OS when you install (or re-configure using YaST).
To clarify once and for all because I’m seeing this is leading to magic weed guesses thus totally deviating from a striaght -and proper- answer.
My only interest is setting this freaking Windows 10 KVM virtual machine on openSUSE host, Windows.
For Linux I already have this hardware crapy almost dying bare metal rig with THIS openSUSE Leap 42.3 installed. If I wanted more Linux boxes, virtual indeed, I’ll certainly better use a Proxmox server.
So trying the questions again,
–do these libvirt’s “current” and “maximum” memory settings really do automatic or “dynamic” memory ballooning?
–So isn’t it possible to set virtual machine’s clock offset to local or UTC? Only with qemu? Windows OSs ship with local clock by default, but I already mentioned what I pretend to attempt in the ArchLinux link reference…
When configuring a VM’s time, you first need to decide what you wish to sync with… The local hardware clock or a network Time Server. The general recommendation is to sync with a network Time Server.
A separate issue is setting your time to the proper time zone. Under the hood, the computer system will do it only one way… UTC. It will define the time zone as the offset from GMT. But, this <single method> can be represented to the User either in the UTC offset or as a human-friendly “local” time zone name. If represented as a local time zone, this is done at the GUI level in the operating system (in your case, Windows) and <not> at some lower level including any setting in the HostOS or lower levels of the Guest. In other words, this setting has nothing to do with virtualization.
Regarding clock, I searched again in all libvirt virtual machine settings and I couldn’t find any clock settings, which is why I asked. Windows installer doesn’t neither have any custom options for clock, it always assumes it set to “local”.
Regarding network, I see a warning icon beside the network source drop menu saying “libvirt version doesn’t support physical device listing”. Yet drop menu has eth0 and wlan0 “macvtap” options, as well as “manually specify a device”, or something like that. How can it list eth0 and wlan0 devices if it “doesn’t support physical device listing”?
Default source mode is “bridge”, with yet another warning saying “in most settings macvtap doesn’t work with host to guest communication”. So Windows virtual machine won’t be able to communicate with openSUSE host nor any PC in physical local network? Won’t it have internet access?
In other miscellaneous stuff, I saw the BIOS setting is just set as “BIOS”, with a warning saying “no UEFI image found in host” (or in hypervisor…). Certainly this laptop’s BIOS is not UEFI. Nothing to worry about?
And finally, in spice monitor settings I see an available checkbox “OpenGL”, with a warning saying “this requires virtio driver with accel3d”. By chance, does anyone has idea about this? Could Windows virtio ISO driver “qxl” be about it? Or better leave this entire setting totally alone?
Although I’m happy to provide answers as I understand (or know), it’s good form to do a little Googling on topics to see if you can find answers on your own first. I’d be the first to admit I’m not as accurate even as many authoritative articles you might find online, and oftentimes articles will be more informative and explain things better than what I might state in a few sentences.
Or, if there is something you suspect is specific to openSUSE that might be different than what you find, asking here would be advisable.
When/if you find a topic that needs further clarification or you can’t find an answer at all, I <hope> that I might then be able to provide an answer.
As I tried to explain, this is a setting that is <in> the Guest, not in virtualization settings.
The VM BIOS will be able to acquire “hardware” time settings from the HostOS physical machine BIOS/UEFI and pass on to the Guest OS, so there isn’t anything to configure. This is all automatic.
So, your GuestOS (no matter if Windows/Linux/other) can then configure Time the same as on physical hardware.
First, you can read a bit about MacVtap. It used to have to be specified and possibly configured on its own, but it’s been a part of the Linux kernel’s support for networking for several versions/years now (approx kernel 3.4?). So, fuggetabout it. If your networking requires MacVtap functionality, it’ll all happen automatically.
I don’t know exactly where the dropdown you’re referencing is, but it’s likely either a naming convention or you’re trying to configure a Guest’s networking to connect to a physical interface instead of a bridging device. If the former, then that’s a simple fix, use interface names that are consistent with the system. If the latter, then you should stop what you’re doing and read some documentation how to create virtual networks and expose them to your Guests. I’ve written numerous posts describing how to do this in this Virtualization and the Networking forums, but I’d rather you just read the docs or other authoritative source. If you <still> can’t figure out how this is supposed to work (after looking at whatever you tried to understand), only then I might put in the time to write up something that should stand the tests of scrutiny and time.
Again, I don’t know if you’re referring to a Guest machine’s virtual BIOS or the real, physical BIOS of the Host machine. If in the Guest, it has nothing to do with the real BIOS (eg you can typically have a virtual legacy BIOS in the Guest while the physical Host machine is UEFI).
In your Guest VM’s properties, check the boxes that enable acceleration.
At least for your Windows Guest, you should have installed virtio drivers.
Then, it’s YMMV.
OpenGL and other intensive graphics support has always been spotty, and this is why for the past few years there has been considerable interest (and slow progress) towards enabling viable GPU pass through. Just don’t expect to be able to run some kind of MMOFPS game and expect good performance.
Just objectively asking, am I really too hard to understand?
Just in case, I’ll elaborate again the network options.
So one starts Create virtual machine application, and goes through until step 5 of 5, where virtual machine settings review is shown and there’s the networking options.
>Network selection
…“Network source” dropdown box to select options (beside this box being the “libvirt no support physical device listing” warning):
…–No virtual networks available (default selected, and it’s grayed out)
…–Host device eth0/wlan0: macvtap
…For any of these 2 options, source modes available (beside this box being the “macvtap no host-guest communication” warning):
…–Bridge
…–VEPA
…–Private
…–“Pasarela” (my system is in Spanish, so what would it be in English? NAT?)
…–Specify shared device name
…For which “Bridge name” is asked
Just to confirm, I did accept the recommended “network bridge” installation (whatever it may be…) back when installing libvirt through Yast. My needs in the Windows virtual machine are for it to be able to communicate with openSUSE host, with other PCs in physical local network, with other Windows virtual machines if I installed more, and internet access. All at same time. Without having to shutdown virtual machine, tweak network settings and starting virtual machine again. Heck, that’s precisely a VirtualBox limitation, which I really didn’t expect in KVM!!
I did read about this macvtap vulgar thing. The VEPA mode requires a physical switch with support for some weird feature still very rare to find nowadays and which name I cannot recall, and bridge mode is designed to NOT work with host-guest communication!!
On Fri 18 Aug 2017 06:56:01 PM CDT, F style wrote:
Just objectively asking, am I really too hard to understand?
Just in case, I’ll elaborate again the network options.
So one starts Create virtual machine application, and goes through until
step 5 of 5, where virtual machine settings review is shown and there’s
the networking options.
>Network selection
…“Network source” dropdown box to select options (beside this box
being the “libvirt no support physical device listing” warning):
…–No virtual networks available (default selected, and it’s grayed
out)
…–Host device eth0/wlan0: macvtap
…For any of these 2 options, source modes available (beside
this box being the “macvtap no host-guest communication” warning):
…–Bridge
…–VEPA
…–Private
…–“Pasarela” (my system is in Spanish, so what would it be
in English? NAT?)
…–Specify shared device name
…For which “Bridge name” is asked
Just to confirm, I did accept the recommended “network bridge”
installation (whatever it may be…) back when installing libvirt
through Yast. My needs in the Windows virtual machine are for it to be
able to communicate with openSUSE host, with other PCs in physical local
network, with other Windows virtual machines if I installed more, and
internet access. All at same time. Without having to shutdown virtual
machine, tweak network settings and starting virtual machine again.
Heck, that’s precisely a VirtualBox limitation, which I really didn’t
expect in KVM!!
I did read about this macvtap vulgar thing. The VEPA mode requires a
physical switch with support for some weird feature still very rare to
find nowadays and which name I cannot recall, and bridge mode is
designed to NOT work with host-guest communication!!
Hi
Unless you have physical additional ethernet ports select NAT
(Pasarela) this will then share you current ethernet port (not
wireless) on the host with your guest(s).
Or you could use wireless for your host machine and then allocate the
ethernet port (NATed) for your VM’s all on the same network.
If you have multiple ethernet ports eg eth0, eth1 and eth2, then you
could on the host create a bridge (eth0=br0, eth1=br1 etc ) and then in
your VM allocate the brN to a virtual machine.
–
Cheers Malcolm °¿° SUSE Knowledge Partner (Linux Counter #276890)
openSUSE Leap 42.2|GNOME 3.20.2|4.4.79-18.26-default
If you find this post helpful and are logged into the web interface,
please show your appreciation and click on the star below… Thanks!
Virt-manager seems to ship somewhat SPICE-ready, by judging default settings for virtual machines “Display Spice” and “Video QXL”. This time I read more about SPICE before asking: https://pve.proxmox.com/wiki/SPICE https://www.linux-kvm.org/page/SPICE
Both SLES and openSUSE KVM docs almost totally lack SPICE information.
Both links advise in general to just set the correct settings for virtual machine’s video, install Windows guest and after that install in it the SPICE client program.
However, and since I’m using the virtio drivers iso for windows, I notice this iso has also qxl drivers. So I became confused, what would be the best way to install SPICE drivers for Windows virtual guest? Installing client after installation or pre-loading qxl driver during installation just like the balloon driver?
Also, in “Display Spice” all defaults seem good, but I’m doubtful about the OpenGL checkbox. It has a warning icon saying “Spice GL requires virtio graphics configured with accel3d”. What does it mean? Is it just advising the already known fact that Windows guest needs proper drivers?
When you use Internet references, always make sure the information is current.
Your Fedora references Fedora 15, which is ancient and is no longer relevant (early architecture), current is Fedora 26.
Is pretty basic
Nothing new required on the HostOS
Follow instructions to enable SPICE in the Guest (Proxmox article looks current)
Use a SPICE viewing app instead of the usual vm manager client.
Generally, whether you see better graphics performance is YMMV, VNC has also improved over the years.
My main use for SPICE is not for graphics, but for implementing USB over ethernet.