January 2020 - Users - libvirt List Archives

Detecting of features of VMs
by Peter Krempa 29 Jan '20

29 Jan '20

Hi, I was asked by Nir and Eyal of the oVirt project on how to detect whether a certain feature is supported by libvirt. As I thought it might be better to document this publically rather than being lost in a private thread I'm posting this to libvirt-users. The specific question will be answered below. --- There are currently two interfaces which allow discovery of libvirts and in turn qemu's capabilities. Both return an XML which describes various aspects of the hosts or VMs capabilities. virConnectGetCapabilities(conn) https://libvirt.org/formatcaps.html The XML returned by this API describes mostly the host itself, the CPU, NUMA topology, security driver support, migration features but also describes guests architectures supported by the guest. In case of the qemu driver the guest section describes some of the aspects and configurations supported by the default qemu binary, the machine types and virtualization types it supports. Additionally the <features> section describes support for some long-existing features such as ACPI, APIC and disk snapshots. As these features are supported for a long time, it can be assumed that they are present (at least in case of the qemu driver) when the second API is present. Now while the virConnectGetCapabilities API provides mostly useful information about the host, the limitation of not being able to report information about specific qemu binaries or machine types lead to introduction. virConnectGetDomainCapabilities(conn, emulator_binary_path, architecture, machine_type, virt_type, flags) https://libvirt.org/formatdomaincaps.html The XML returned by this API describes the capabilities of an VM which would be started using the parameters given to the API (note that if the parameters are NULL a default is provided). The <os> subelement describes supported OS loader types and firmware binaries. <cpu> describes which CPU models are supported by the emulator and also the CPU which would be used if 'host-model' is selected as the CPU model. The <devices> section describes models and some properties of devices supported by the VM. The 'enum's reported in this XML map to supported values of some of the fields we report the possible configurations for. Note that these values are reported individually and all combinations aren't necessarily supported. The last section 'features' describes whether some high-level features supported by such a VM. The features themselves are described in the documentation above. Note that the entries into 'features' are added arbitrarily depending on whether a feature is deemed important enough to be exposed. So please feel free to report if anything is missing or it would be helpful if we reported any feature. --- Now specifically to the question: I was asked on when it's safe to pass VIR_DOMAIN_UNDEFINE_CHECKPOINTS_METADATA to virDomainUndefine. In this case it's safe to pass it if <features> contains the <backup> element regardless of the value of 'supported' attribute. I'll also mention this in the documentation of the domain caps XML.

2 1

Host CPU usage high on guest network load (macvtap).
by André Malm 28 Jan '20

28 Jan '20

Hello, Is it reasonable for one CPU on the host to be loaded 20-30% on 100 mbps download in a guest? There is barely any (1-2%) load in the guest. The host is running ubuntu 18.04 on an Intel i7-7700K, virtualization etc is enabled. Everything except host load on guest network activity is fine. This is how i setup the network with virt-install: --network type=direct,source=enp1s0,source_mode=bridge,model=virtio,mac=00:11:22:33:44:55 In the guest (ubuntu18.04 cloud minimal) i run a simple wget, setting mode to vepa still works but the cpu load is the same.

1 0

Emulation packages for Centos8
by Mauricio Tavares 28 Jan '20

28 Jan '20

So I was looking for the qemu-system-ppc so I could emulate that cpu in my Intel I7 centos8 box. So far I found that available for fedora (and ubuntu) but not centos. Does anyone have a suggestion for where (repo) I can get said package?

1 0

NFS and unsafe migration
by Gionatan Danti 28 Jan '20

28 Jan '20

Hi all, I have a question about NFS datastore and unsafe migration. When migrating a virtual machine having a virtio disk in writeback cache between two hosts sharing a single NFS datastore, I get the following error: "Unsafe migration: Migration may lead to data corruption if disks use cache != none or cache != directsync" I understand why libvirt alerts for unsafe migration in cases where no coherency is enfoced by the underlying system; however, is it really the case for nfs? From what I know (and from the man page), by default nfs has open-to-close consistency, which seems quite right for migrating a virtual machine between (as only one host at a time reads/writes/locks the virtual disk files). I know that I can simply use cache=none to let the problem go away; however, this significantly impairs performance when used on nfs. Am I missing something? Thanks. -- 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

1 0

KVM not available on system bus
by Kasper Laudrup 27 Jan '20

27 Jan '20

Hi libvirt-users, Hope this is the right place to ask, otherwise please point me in the right direction. I have a libvirt virtual machine running on the session bus that I would like to access through SSH. I have previously done so using X11 forwarding and while it works, it is very sluggish with the connection I have. I recently learned that you can access the virtual machine with virt-viewer through SSH which should perform much better. Unfortunately, my virtual machine is currently running on the session bus and that doesn't seem to be supported (please do correct if I'm wrong here). It doesn't really matter to much if the VM is running on the system bus or session bus, I just prefer running things as a non privileged user for all the obvious reasons, so that's where I created the VM initially. I tried to follow this guide to move my VM to the system bus. Running `sudo virsh define vm.xml` fails with: error: invalid argument: could not find capabilities for arch=x86_64 domaintype=kvm Digging a bit further into it, I figured out that the cause of the error message is, that I for some reason do not have KVM acceleration support when running VMs on the system bus (as root) running my VM on the session bus as a normal user (with the correct group membership) works fine. Trying to launch virt-manager as root verifies that, as creating a new VM warns me that I do not have KVM support. I'm fairly lost as to what to do from here. I must admit I remember struggling a bit to get the virtual machine to run with KVM support on the session bus in the first place, but have completely forgotten what the problem and resolution was. I'm using Debian stable (Buster) with standard package versions: QEMU 3.1.0 libvirt 5.0.0 Any kind of help or input would be greatly appreciated. Thanks a lot and kind regards, Kasper Laudrup

2 2

VM crash, partitiontable deleted CPU 1/KVM[2867866]: segfault at 2 ip 0000560cffe03757 sp 00007f0e8babaeb0 error 4 in qemu-system-x86_64
by Oliver Dzombic 27 Jan '20

27 Jan '20

Hi, Got from a running machine suddenly: [859069.022739] CPU 1/KVM[2867866]: segfault at 2 ip 0000560cffe03757 sp 00007f0e8babaeb0 error 4 in qemu-system-x86_64[560cffbd4000+45d000] [859069.022749] Code: 00 00 00 41 57 41 56 41 55 41 54 49 89 cc b9 a2 03 00 00 55 48 89 f5 48 8d 35 1c 35 29 00 53 48 89 fb 48 83 c7 38 48 83 ec 38 <44> 0f b7 52 02 4c 8b 7a 08 4c 89 44 24 08 4c 8d 05 b4 08 29 00 45 [859087.661975] device k1806-YcXLF left promiscuous mode # qemu-system-x86_64 --version QEMU emulator version 4.2.0 (qemu-4.2.0-2.fc31) Copyright (c) 2003-2019 Fabrice Bellard and the QEMU Project developers # virsh -V Virsh command line tool of libvirt 6.0.0 See web site at https://libvirt.org/ Compiled with support for: Hypervisors: QEMU/KVM LXC LibXL OpenVZ VMware VirtualBox ESX Hyper-V Test Networking: Remote Network Bridging Interface netcf Nwfilter VirtualPort Storage: Dir Disk Filesystem SCSI Multipath iSCSI LVM RBD Sheepdog Gluster ZFS Miscellaneous: Daemon Nodedev SELinux Secrets Debug DTrace Readline After that the partition table of the zfs volume was cleared. I also opened a bug report with all informations i could get so far: https://bugzilla.redhat.com/show_bug.cgi?id=1795082 I am thankful for any idea / suggestion. Thank you -- Mit freundlichen Gruessen / Best regards Oliver Dzombic Layer7 Networks mailto:info@layer7.net Anschrift: Layer7 Networks GmbH Zum Sonnenberg 1-3 63571 Gelnhausen HRB 96293 beim Amtsgericht Hanau Geschäftsführung: Oliver Dzombic UST ID: DE259845632

1 0

VM crash, partitiontable deleted CPU 1/KVM[2867866]: segfault at 2 ip 0000560cffe03757 sp 00007f0e8babaeb0 error 4 in qemu-system-x86_64
by Oliver Dzombic 27 Jan '20

27 Jan '20

Hi, Got from a running machine suddenly: [859069.022739] CPU 1/KVM[2867866]: segfault at 2 ip 0000560cffe03757 sp 00007f0e8babaeb0 error 4 in qemu-system-x86_64[560cffbd4000+45d000] [859069.022749] Code: 00 00 00 41 57 41 56 41 55 41 54 49 89 cc b9 a2 03 00 00 55 48 89 f5 48 8d 35 1c 35 29 00 53 48 89 fb 48 83 c7 38 48 83 ec 38 <44> 0f b7 52 02 4c 8b 7a 08 4c 89 44 24 08 4c 8d 05 b4 08 29 00 45 [859087.661975] device k1806-YcXLF left promiscuous mode # qemu-system-x86_64 --version QEMU emulator version 4.2.0 (qemu-4.2.0-2.fc31) Copyright (c) 2003-2019 Fabrice Bellard and the QEMU Project developers # virsh -V Virsh command line tool of libvirt 6.0.0 See web site at https://libvirt.org/ Compiled with support for: Hypervisors: QEMU/KVM LXC LibXL OpenVZ VMware VirtualBox ESX Hyper-V Test Networking: Remote Network Bridging Interface netcf Nwfilter VirtualPort Storage: Dir Disk Filesystem SCSI Multipath iSCSI LVM RBD Sheepdog Gluster ZFS Miscellaneous: Daemon Nodedev SELinux Secrets Debug DTrace Readline After that the partition table of the zfs volume was cleared. I also opened a bug report with all informations i could get so far: https://bugzilla.redhat.com/show_bug.cgi?id=1795082 I am thankful for any idea / suggestion. Thank you ! -- Mit freundlichen Gruessen / Best regards Oliver Dzombic Layer7 Networks mailto:info@layer7.net Anschrift: Layer7 Networks GmbH Zum Sonnenberg 1-3 63571 Gelnhausen HRB 96293 beim Amtsgericht Hanau Geschäftsführung: Oliver Dzombic UST ID: DE259845632

1 0

Prevent the firewall from being compromised through libvirtd
by Thomas Luening 27 Jan '20

27 Jan '20

Hello @ all The libvirt-daemon compromises the packet-filtering-rules at daemon-startup, before any VM is started. To prevent this, I first have create a hook-script which deletes existing rules, but apparently these rules are set after the hook. Removing the defined networks was no solution either. Worst of all is, a service restart of the daemon may even completely neutralize the firewall. Is there a solution to prevent this undesirable behavior? No matter how or who what do or with what network configuration a VM is started, the daemon must not compromise the firewall, by altering them. The Firewall is untouchable and taboo. What can I do to disable that? Thank you! Best Regards Tom $ dpkg -l libvirt-daemon ||/ Name Version Architektur Beschreibung +++-=========================-============-============-================================== ii libvirt-daemon 5.0.0-4 amd64 Virtualization daemon $ lsb_release -a Distributor ID: Debian Description: Debian GNU/Linux 10 (buster) Release: 10 Codename: buster

2 1

Repetitive RBD disk definitions
by wferi＠niif.hu 24 Jan '20

24 Jan '20

Hi, I'm using libvirt 5.0.0 on a Ceph cluster. The VM disks are all from the same Ceph pool, so all my <disk> elements are basically the same: <disk type="network" device="disk"> <driver name="qemu" type="raw" cache="none"/> <source protocol="rbd" name="vmdisks/VOLNAME"> <host name="x.y.z.1" port="6789"/> <host name="x.y.z.2" port="6789"/> <host name="x.y.z.3" port="6789"/> </source> <auth username="libvirt"> <secret type="ceph" usage="libvirt_key"/> </auth> <target dev="vdX"/> </disk> ... apart from the VOLNAME and vdX words. Is there a way to factor out (at least part of) the common stuff? This repetition isn't totally unmanageable, just not pretty. Creating a libvirt storage pool seemed like a good solution at first, but I couldn't find a way to "instantiate" volumes in the domain XML (and defining them cluster-wide would mean duplicating what Ceph does, which I'd rather avoid). -- I'd be grateful for any ideas, Feri.

1 0

virsh vol-download uses a lot of memory
by R. Diez 23 Jan '20

23 Jan '20

Hi all: I am using the libvirt version that comes with Ubuntu 18.04.3 LTS. I have written a script that backs up my virtual machines every night. I want to limit the amount of memory that this backup operation consumes, mainly to prevent page cache thrashing. I have described the Linux page cache thrashing issue in detail here: http://rdiez.shoutwiki.com/wiki/Today%27s_Operating_Systems_are_still_incre… The VM virtual disk weighs 140 GB at the moment. I thought 500 MiB of RAM should be more than enough to back it up, so I added the following options to the systemd service file associated to the systemd timer I am using: MemoryLimit=500M However, the OOM is killing "virsh vol-download": Jan 21 23:40:00 GS-CEL-L kernel: [55535.913525] [ pid ] uid tgid total_vm rss pgtables_bytes swapents oom_score_adj name Jan 21 23:40:00 GS-CEL-L kernel: [55535.913527] [ 13232] 1000 13232 5030 786 77824 103 0 BackupWindows10 Jan 21 23:40:00 GS-CEL-L kernel: [55535.913528] [ 13267] 1000 13267 5063 567 73728 132 0 BackupWindows10 Jan 21 23:40:00 GS-CEL-L kernel: [55535.913529] [ 13421] 1000 13421 5063 458 73728 132 0 BackupWindows10 Jan 21 23:40:00 GS-CEL-L kernel: [55535.913530] [ 13428] 1000 13428 712847 124686 5586944 523997 0 virsh Jan 21 23:40:00 GS-CEL-L kernel: [55535.913532] oom-kill:constraint=CONSTRAINT_MEMCG,nodemask=(null),cpuset=/,mems_allowed=0,oom_memcg=/system.slice/VmBackup.service,task_memcg=/system.slice/VmBackup.service,task=virsh,pid=13428,uid=1000 Jan 21 23:40:00 GS-CEL-L kernel: [55535.913538] Memory cgroup out of memory: Killed process 13428 (virsh) total-vm:2851388kB, anon-rss:486180kB, file-rss:12564kB, shmem-rss:0kB I wonder why "virsh vol-download" needs so much RAM. It does not get killed straight away, it takes a few minutes to get killed. It starts using a VMSIZE of around 295 MiB, which is not really frugal for a file download operation, but then it grows and grows. Note that the virtual machine is not running (shut off) while doing the backup. Last time I tried with an increased memory limit of 5G, "virsh vol-download" was killed when using 7,4 G, and the partially-downloaded volume file weighted 60G. Therefore, it looks like "virsh vol-download" is using a percentage of the downloaded size in RAM. Is there a way to make "virsh vol-download" use less memory? Thanks in advance, rdiez

4 8