- Users - libvirt List Archives

Re: Predictable and consistent net interface naming in guests
by Igor Mammedov 03 Nov '22

03 Nov '22

On Thu, 3 Nov 2022 00:13:16 +0200 Amnon Ilan <ailan(a)redhat.com> wrote: > On Wed, Nov 2, 2022 at 6:47 PM Laine Stump <laine(a)redhat.com> wrote: > > > On 11/2/22 11:58 AM, Igor Mammedov wrote: > > > On Wed, 2 Nov 2022 15:20:39 +0000 > > > Daniel P. Berrangé <berrange(a)redhat.com> wrote: > > > > > >> On Wed, Nov 02, 2022 at 04:08:43PM +0100, Igor Mammedov wrote: > > >>> On Wed, 2 Nov 2022 10:43:10 -0400 > > >>> Laine Stump <laine(a)redhat.com> wrote: > > >>> > > >>>> On 11/1/22 7:46 AM, Igor Mammedov wrote: > > >>>>> On Mon, 31 Oct 2022 14:48:54 +0000 > > >>>>> Daniel P. Berrangé <berrange(a)redhat.com> wrote: > > >>>>> > > >>>>>> On Mon, Oct 31, 2022 at 04:32:27PM +0200, Edward Haas wrote: > > >>>>>>> Hi Igor and Laine, > > >>>>>>> > > >>>>>>> I would like to revive a 2 years old discussion [1] about > > consistent network > > >>>>>>> interfaces in the guest. > > >>>>>>> > > >>>>>>> That discussion mentioned that a guest PCI address may change in > > two cases: > > >>>>>>> - The PCI topology changes. > > >>>>>>> - The machine type changes. > > >>>>>>> > > >>>>>>> Usually, the machine type is not expected to change, especially if > > one > > >>>>>>> wants to allow migrations between nodes. > > >>>>>>> I would hope to argue this should not be problematic in practice, > > because > > >>>>>>> guest images would be made per a specific machine type. > > >>>>>>> > > >>>>>>> Regarding the PCI topology, I am not sure I understand what changes > > >>>>>>> need to occur to the domxml for a defined guest PCI address to > > change. > > >>>>>>> The only think that I can think of is a scenario where > > hotplug/unplug is > > >>>>>>> used, > > >>>>>>> but even then I would expect existing devices to preserve their > > PCI address > > >>>>>>> and the plug/unplug device to have a reserved address managed by > > the one > > >>>>>>> acting on it (the management system). > > >>>>>>> > > >>>>>>> Could you please help clarify in which scenarios the PCI topology > > can cause > > >>>>>>> a mess to the naming of interfaces in the guest? > > >>>>>>> > > >>>>>>> Are there any plans to add the acpi_index support? > > >>>>>> > > >>>>>> This was implemented a year & a half ago > > >>>>>> > > >>>>>> https://libvirt.org/formatdomain.html#network-interfaces > > >>>>>> > > >>>>>> though due to QEMU limitations this only works for the old > > >>>>>> i440fx chipset, not Q35 yet. > > >>>>> > > >>>>> Q35 should work partially too. In its case acpi-index support > > >>>>> is limited to hotplug enabled root-ports and PCIe-PCI bridges. > > >>>>> One also has to enable ACPI PCI hotplug (it's enled by default > > >>>>> on recent machine types) for it to work (i.e.it's not supported > > >>>>> in native PCIe hotplug mode). > > >>>>> > > >>>>> So if mgmt can put nics on root-ports/bridges, then acpi-index > > >>>>> should just work on Q35 as well. > > >>>> > > >>>> With only a few exceptions (e.g. the first ich9 audio device, which is > > >>>> placed directly on the root bus at 00:1B.0 because that is where the > > >>>> ich9 audio device is located on actual Q35 hardware), libvirt will > > >>>> automatically put all PCI devices (including network interfaces) on a > > >>>> pcie-root-port. > > >>>> > > >>>> After seeing reports that "acpi index doesn't work with Q35 > > >>>> machinetypes" I just assumed that was correct and didn't try it. But > > >>>> after seeing the "should work partially" statement above, I tried it > > >>>> just now and an <interface> of a Q35 guest that had its PCI address > > >>>> auto-assigned by libvirt (and so was placed on a pcie-root-port)m and > > >>>> had <acpi index='4'/> was given the name "eno4". So what exactly is it > > >>>> that *doesn't* work? > > >>> > > >>> From QEMU side: > > >>> acpi-index requires: > > >>> 1. acpi pci hotplug enabled (which is default on relatively new q35 > > machine types) > > >>> 2. hotpluggble pci bus (root-port, various pci bridges) > > >>> 3. NIC can be cold or hotplugged, guest should pick up acpi-index of > > the device > > >>> currently plugged into slot > > >>> what doesn't work: > > >>> 1. device attached to host-bridge directly (work in progress) > > >>> (q35) > > >>> 2. devices attached to any PXB port and any hierarchy hanging of it > > (there are not plans to make it work) > > >>> (q35, pc) > > >> > > >> I'd say this is still a relatively important, as the PXBs are needed > > >> to create a NUMA placement aware topology for guests, and I'd say it > > >> is undesirable to loose acpi-index if a guest is updated to be NUMA > > >> aware, or if a guest image can be deployed in either normal or NUMA > > >> aware setups. ... > How big of a project would it be to enable ACPI-indexing/hotplug with PXB? > Since native PCI was improved, we can still compromise on switching to > native-PCI-hotplug when PXB is required (and no fixed indexing) My guesstimate would be it's not terribly difficult. Maybe we could even marry native hotplug & acpi-index after the later is decoupled from ACPI PCI hotplug as much as possible. > > Thanks, > Amnon > > > > > > > Anyway, it sounds like (*within the confines of how libvirt constructs > > the PCI topology*) we actually have functional parity of acpi-index > > between 440fx and Q35. > > > >

1 0

Predictable and consistent net interface naming in guests
by Edward Haas 03 Nov '22

03 Nov '22

Hi Igor and Laine, I would like to revive a 2 years old discussion [1] about consistent network interfaces in the guest. That discussion mentioned that a guest PCI address may change in two cases: - The PCI topology changes. - The machine type changes. Usually, the machine type is not expected to change, especially if one wants to allow migrations between nodes. I would hope to argue this should not be problematic in practice, because guest images would be made per a specific machine type. Regarding the PCI topology, I am not sure I understand what changes need to occur to the domxml for a defined guest PCI address to change. The only think that I can think of is a scenario where hotplug/unplug is used, but even then I would expect existing devices to preserve their PCI address and the plug/unplug device to have a reserved address managed by the one acting on it (the management system). Could you please help clarify in which scenarios the PCI topology can cause a mess to the naming of interfaces in the guest? Are there any plans to add the acpi_index support? [1] https://bugzilla.redhat.com/show_bug.cgi?id=1874096#c15 Thank you, Edy.

6 13

DOS
by Tomas By 02 Nov '22

02 Nov '22

Hi all, I am trying to set up an `isolated virtual network' with two or more MS-DOS guests, all on one single Linux box. The aim is for them to share a disk (or just one directory on one disk). As I could not get it to work with IPX, am now trying TCP/IP. However, I suspect the MS tools are still attempting to use IPX as well. Is there some simple solution that I am missing? Does libvirt support IPX at all? Any other easy way to do DOS file sharing? (I believe I can use any version of DOS, and any sort of tool as long as I get a `remote' directory available as some drive letter.) /Tomas

2 4

hook script for changing disk definition before guest start
by Gionatan Danti 01 Nov '22

01 Nov '22

Hi all, due to this [1] I am trying to change a guest's disk definition from "pool volume" to "file" via hook scripts. This seems an easy task via virsh detach-disk and virsh attach-disk, but virsh can not be called inside libvirt hook scripts. The best approach would be to change the disk definition right when a disk is attached, but I can not find any event calling a hook script when a disk is attache/detached. Do you have any suggestion short of monitoring the directory /etc/libvirt/qemu itself for changed files? Thanks. [1] https://github.com/cockpit-project/cockpit-machines/issues/815 -- Danti Gionatan Supporto Tecnico Assyoma S.r.l. - www.assyoma.it email: g.danti(a)assyoma.it - info(a)assyoma.it GPG public key ID: FF5F32A8

2 3

Re: How to check tap device for sndbuf when I set sndbuf=0 in the interface xml
by Yalan Zhang 31 Oct '22

31 Oct '22

cc libvirt-users for more inputs Hi, I'm trying to find out how to check sndbuf for a tap device, could you please help to check it? In kernel 2.6.18, the tap device's default sndbuf is 1MB. There is a RFE bug[1] and patch[2] to introduce the option below in libvirt to adjust the sndbuf. <tune> <sndbuf>1600</sndbuf> </tune> It is said in the patch[2] that when we set sndbuf=0, we actually set it to 0xffffffff. How to check if it is set successfully? I have checked the below files, the value didn't change after I start a vm with sndbuf=0 # grep . /proc/sys/net/core/*mem_default /proc/sys/net/core/rmem_default:212992 /proc/sys/net/core/wmem_default:212992 And from the doc[3]: "The default value is set by the /proc/sys/net/core/wmem_default file and the maximum allowed value is set by the /proc/sys/net/core/wmem_max file." Current kernel 5.14.0-177.el9 has some updates about sndbuf: # cat /proc/sys/net/core/wmem_default 212992 # cat /proc/sys/net/core/wmem_max 212992 Thank you! [1] Bug 665293 <https://bugzilla.redhat.com/show_bug.cgi?id=665293> - RFE: Allow setting size of send buffer per TAP device in QEMU driver [2] https://listman.redhat.com/archives/libvir-list/2011-January/032763.html [3] https://man7.org/linux/man-pages/man7/socket.7.html Yalan

1 0

Video device model set resolution does not work
by 陈新隆 31 Oct '22

31 Oct '22

Hi team, I've add a resolution sub-element to vga video device model, and the x attribute is 1920, y attribute is 1080. Dumped domain xml as follow : bash-4.4# virsh list > Authorization not available. Check if polkit service is running or see > debug message for more information. > Id Name State > ----------------------------------- > 1 proj1_vm-4uaz57bh running > bash-4.4# virsh dumpxml proj1_vm-4uaz57bh | grep -i -10 vga > Authorization not available. Check if polkit service is running or see > debug message for more information. > <alias name='input1'/> > </input> > <input type='keyboard' bus='ps2'> > <alias name='input2'/> > </input> > <graphics type='vnc' > socket='/var/run/kubevirt-private/dfe7d514-63f1-4e97-b642-f93109d5aa2f/virt-vnc'> > <listen type='socket' > socket='/var/run/kubevirt-private/dfe7d514-63f1-4e97-b642-f93109d5aa2f/virt-vnc'/> > </graphics> > <audio id='1' type='none'/> > <video> > <model type='vga' vram='65536' heads='1' primary='yes'> > <resolution x='1920' y='1080'/> > </model> > <alias name='video0'/> > <address type='pci' domain='0x0000' bus='0x00' slot='0x01' > function='0x0'/> > </video> > <memballoon model='virtio-non-transitional'> > <stats period='10'/> > <alias name='balloon0'/> > <address type='pci' domain='0x0000' bus='0x07' slot='0x00' > function='0x0'/> > </memballoon> It does work on my ubuntu guest, but it doesn't work on centos / windows guest. Is there any way to solve this problem ?

1 0

Booting from USB device
by Simon Fairweather 28 Oct '22

28 Oct '22

I have a hooks file to hotplug a USB device at the prepare stage. Using OVMF version 2.6 this works fine, but using OVMF 2.7 the guest no longer boots from the USB device. Option 1) For it to work I have to set the boot order on the USB device and once the OVMF Shell appears I can reset it and it will find the USB device and boot. Option 2) Another setup which works if I remove all boot order settings from the guest and have a boot dev = 'fd' and this works every time.but I cannot add multiple entries. <os> <type arch='x86_64' machine='pc-q35-5.1'>hvm</type> <loader readonly='yes' type='pflash'>/usr/share/qemu/ovmf-x64/OVMF_CODE-pure-efi-tpm.fd</loader> <nvram>/etc/libvirt/qemu/nvram/b96bfc99-3612-0daa-9d8c-69e9d550d05a_VARS-pure-efi-tpm.fd</nvram> <boot dev='fd'/> <boot dev='hd'/> Any thoughts on how to get option 1) to working without having to do a reset.

1 0

failed to load seccomp syscall filter in kernel: Operation canceled
by Jiatong Shen 28 Oct '22

28 Oct '22

Hello community, I am operating an openstack cluster where applications (libvirt/nova etc) are running using containers. The compute node's arch is aarch64 (phytium 2500), when there are virtual machines around 60 or 70, I failed to boot new virtual machines and faced with following error message, error: internal error: qemu unexpectedly closed the monitor: 2022-10-24T06:23:54.545685Z qemu-system-aarch64: -sandbox on,obsolete=deny,elevateprivileges=deny,spawn=deny,resourcecontrol=deny: failed to load seccomp syscall filter in kernel: Operation canceled Interestingly, if I virsh stop one virtual machine, I am able to boot another. Besides, I managed to manually boot a virtual machine without any issue. So my question is what could be the potential cause of this behavior and how can I deal with it? Thank you very much in advance for the help. -- Best Regards, Jiatong Shen

2 2

Virtual Networking
by Armin Lepir 26 Oct '22

26 Oct '22

Hi I am using libvrt for the first time. Im building a KVM for multiple Virtual OS instances. The problem i have is with your official documentation for Virtual Networking. https://wiki.libvirt.org/page/VirtualNetworking The following is wrong: The default mode is BRIDGE + NAT. Optional mode is ROUTING. It should be: The default mode is ROUTING + NAT. Optional mode is BRIDGE. As far as i know a bridge operates on the Layer2. IP and NAT operate on the Layer3. NAT can not operate on the L2. Please tell me that im wrong and explain how am i wrong. Best Regards

2 2

Re: MemoryBacking options
by Michal Prívozník 24 Oct '22

24 Oct '22

[Once again, I ask you to keep the list on the CC. In gmail's web UI it's the 'Reply to all' button]. On 10/22/22 11:19, Simon Fairweather wrote: > Is there a way to limit the shared memory allocated via Libvirt to > memory backing? > Yes, there is. On the top level there's /domain/memoryBacking [1] that serves as the default value for the guest. Then, under /domain/cpu/numa/cell [2] individual guest NUMA nodes are configured and each can have @memAccess attribute which overrides the top level setting (=memoryBacking). And finally, under /domain/devices/memory [3] individual RAM modules can be configured, and each one can have @access attribute which overrides both top level AND guest NUMA node setting. For instance: <domain> <maxMemory slots='16' unit='GiB'>8</maxMemory> <memory unit='GiB'>5</memory> <memoryBacking> <access mode='private'/> </memoryBacking> ... <cpu> <numa> <cell id='0' memory='2' unit='GiB' memAccess='shared'/> <cell id='1' memory='2' unit='GiB' /> </numa> </cpu> ... <devices> <memory model='dimm' access='private'> <target> <size unit='GiB'>1</size> <node>0</node> </target> </memory> </device> </domain> Here, the guest NUMA node 0 is going to be shared, except for DIMM module attached to it (which is going to be private) and NUMA node 1 is going to be private again. Michal 1: https://libvirt.org/formatdomain.html#memory-backing 2: https://libvirt.org/formatdomain.html#cpu-model-and-topology 3: https://libvirt.org/formatdomain.html#memory-devices > > > On Mon, Oct 17, 2022 at 9:06 AM Michal Prívozník <mprivozn(a)redhat.com > <mailto:mprivozn@redhat.com>> wrote: > > [please keep the list on CC for benefit of the community] > On 10/16/22 09:58, Simon Fairweather wrote: > > Thanks Michal, Are there any known issues with Memory Backing. We have > > QEMU 7.1 and libvirt 8.7. > > > > If just Memory Backing is set the VM Freezes after a couple of > hours and > > the allocated CPUs show 100% on the host. > > > > Seems to happen more on AMD Platforms than Intel. > > > > Please advise on any info you need, likely this is a QEMU issue. > > > > <memoryBacking> > > <source type='memfd'/> > > <access mode='shared'/> > > </memoryBacking> > > Agreed, this smells like QEMU issue. There's nothing obviously wrong > with this snipped. I have a VM with just that and run it just fine. > Although, I run it for couple of hours max. > > Michal >

1 0