- Users - libvirt List Archives

[libvirt-users] GUEST Memory statistics secret revealed ...
by Jean-Pierre Ribeauville 30 Oct '15

30 Oct '15

Hi, I'm wondering how to explain following metrics I got from libvirt, regarding memory stats for the two GUESTS running on my host. 1) virDomainGetXMLDesc() shows these values : memmax (1024000.000000) memcurrent(1024000.000000) memmax (2048000.000000) .memcurrent(2048000.000000) 2) As values returned by virDomainMemoryStats(), I got these values : VIR_DOMAIN_MEMORY_STAT_ACTUAL_BALLOON: active_balloon (1024000.000000) active_balloon (2048000.000000) VIR_DOMAIN_MEMORY_STAT_RSS: rss(1194195.312500) rss(2039269.531250 3) By using cgroup counters "memory.usage_in_bytes" for both Guests, I got these values : memusage (1 207 207 031 250) memusage (2 834 503 906 250) How is it possible to have a balloon size equal to memcurrent ? What "memory.usage_in_bytes" stands for (it's the maximum value of the three metrics I got for each GUEST) ? FYI , I use a 1.2.8 libvirt. Thanks. Reagrds, J.P. Ribeauville P: +33.(0).1.47.17.20.49 . Puteaux 3 Etage 5 Bureau 4 jpribeauville(a)axway.com<mailto:jpribeauville@axway.com> http://www.axway.com<http://www.axway.com/> P Pensez à l'environnement avant d'imprimer.

1 0

[libvirt-users] failed to pivot job for disk vdc
by Nerijus Baliunas 30 Oct '15

30 Oct '15

Hello, I did a virsh snapshot-create-as --domain meta sn1 --diskspec vda,file=/var/lib/libvirt/images/meta-sn1.qcow2 --disk-only --atomic --no-metadata Then I successfully blockcommitted 2 disks: virsh blockcommit meta vda --active --verbose --pivot virsh blockcommit meta vdb --active --verbose --pivot But when doing virsh blockcommit meta vdc --active --verbose --pivot I got: Block commit: [100 %]error: failed to pivot job for disk vdc error: block copy still active: disk 'vdc' not ready for pivot yet Domain snapshot sn1 created And now it looks like this: virsh domblklist meta Target Source ------------------------------------------------ vda /var/lib/libvirt/images/meta.qcow2 vdb /var/lib/libvirt/images/meta-home.qcow2 vdc /var/lib/libvirt/images/meta-pgsql.sn1 How do I make vdc to be the initial disk, not the snapshot one? Regards, Nerijus

1 4

[libvirt-users] How to retrieve legacy cgroups location ?
by Jean-Pierre Ribeauville 29 Oct '15

29 Oct '15

Hi, As told in "Control Groups Resource Management" libvirt page : Legacy cgroups layout Prior to libvirt 1.0.5, the cgroups layout created by libvirt was different from that described above, and did not allow for administrator customization. Libvirt used a fixed, 3-level hierarchy libvirt/{qemu,lxc}/$VMNAME which was rooted at the point in the hierarchy where libvirtd itself was located. So if libvirtd was placed at /system/libvirtd.service by systemd, the groups for each virtual machine / container would be located at /system/libvirtd.service/libvirt/{qemu,lxc}/$VMNAME. In addition to this, the QEMU drivers further child groups for each vCPU thread and the emulator thread(s). This leads to a hierarchy that looked like I'm trying to retrieve this layout : $ROOT | +- system | +- libvirtd.service | +- libvirt | +- qemu | | | +- vm1 | | | | | +- emulator | | +- vcpu0 | | +- vcpu1 | | | +- vm2 | | | | | +- emulator | | +- vcpu0 | | +- vcpu1 | | | +- vm3 | | | +- emulator | +- vcpu0 | +- vcpu1 How may I find where systemd has placed libvirtd ? I use libvirt 0.10.2 on a RHEL6 2.6.32-431.el6.x86_64 Thx for help. J.P. Ribeauville P: +33.(0).1.47.17.20.49 . Puteaux 3 Etage 5 Bureau 4 jpribeauville(a)axway.com<mailto:jpribeauville@axway.com> http://www.axway.com<http://www.axway.com/> P Pensez à l'environnement avant d'imprimer.

2 2

[libvirt-users] virsh can't support VM offline blockcommit
by justlibvirt＠gmail.com 29 Oct '15

29 Oct '15

Hi everyone! I use the libvirt(version: 1.2.2) and QEMU(version: 2.2.1) to test qemu snapshot features: I tried virsh blockcommit when VM offline, the virsh blockcommit failed: the error messase as below: error: Requested operation is not valid: domain is not running when I start the VM, the virsh blockcommit work fine! my question is：we need virsh support blockcommit when VM offline。how do？ we test qemu snapshot as follows： we need create snapshot of the cycle，and by virsh blockcommit merge snapshot chain，but when VM offline， the virsh blockcommit will fail， someone suggest me：when VM offline，may use qemu-img commit to submit snapshot turn，but qemu-img can't modify snapshot chain backing relation，it need we use qemu-img rebase amend snapshot‘s backingfile， the main question is when we use qemu-img rebase modify snapshot's backingfile, if someone start the VM, it will may lead to VM filesystem readonly。 we should how to solve the question? thanks! justlibvirt(a)gmail.com

2 3

[libvirt-users] How to find memory utilization percent of kvm guest?
by suyog jadhav 28 Oct '15

28 Oct '15

I currently use following approach but its failing Find Max memory using following API vdgiRet = virDomainGetInfo(domPtr, infoPtr); memTotal = infoPtr->memory; Find current utilization using following API rVal = virDomainMemoryStats (domPtr, memStats, VIR_DOMAIN_MEMORY_STAT_NR, 0) ; for (i = 0; i < rVal; i++) { if (memStats[i].tag == 7) memUsed = memStats[i].val; } Which is the value of "VIR_DOMAIN_MEMORY_STAT_RSS = 7" Then I calculate util as (memUsed/memTotal) But the value of memUsed exceeds value of memTotal by small amount resulting in memUtil to be around 101 %. I understand that VIR_DOMAIN_MEMORY_STAT_RSS is Resident Set Size of the process running the domain. How do I Calculate the current memory utilization using alternate way?

1 0

[libvirt-users] How can I set the advertised value of MTU for virbr0 interface?
by George Trakatelis 26 Oct '15

26 Oct '15

Hi all, I want to set the advertised value of MTU for virbr0 interface on a Fedora 21 host to 1454 to accommodate the installation of new VMs for testing purposes. The aforementioned host is an OpenStack VM actually, which explains the need for a lower than 1500 MTU. I noticed Pieter Hollants' patch to set dhcp-option=<number>,"<value>" for an interface, but it seems that it has not been implemented yet. Is there a possibility to have such an option? George

3 3

[libvirt-users] net.ipv4.conf.all.forwarding
by lejeczek 22 Oct '15

22 Oct '15

hi everybody I'd like to ask how libvirtd influences net.ipv4.conf.all.forwarding, would you know? I've noticed that if I use host's bridge, therefor there is no <forward mode= > in my network then "net.ipv4.conf.all.forwarding" goes back to 0. Is this intended I wonder? Looks like libvirt decides it's user responsibility now. many thanks

2 1

[libvirt-users] openvz driver
by Volker Janzen 22 Oct '15

22 Oct '15

Hi, I've a question regarding the openvz driver. I noticed that the virsh dumpxml output does not contain the memory settings. E.g.: <domain type='openvz'> <name>104</name> <uuid>ad2e9d0b-d242-482a-8640-5102e49d0f9b</uuid> <memory unit='KiB'>0</memory> <currentMemory unit='KiB'>0</currentMemory> <vcpu placement='static'>2</vcpu> <os> <type>exe</type> <init>/sbin/init</init> </os> <clock offset='utc'/> <on_poweroff>destroy</on_poweroff> <on_reboot>destroy</on_reboot> <on_crash>destroy</on_crash> <devices> </devices> </domain> Can someone explain why this data is missing? As far as I know the driver reads the container configuration files, which contains the missing information. Regards Volker Viele Grüße Volker

2 1

[libvirt-users] noVNC with qemu over ssh
by Digimer 22 Oct '15

22 Oct '15

Hi all, I have a little webserver interface for managing VMs running on a RHEL 6 based HA cluster hosting KVM/qemu VMs. I'd like to use noVNC to provide access to the hosted VMs, but I am not finding much information on how to do this (or rather, I can't find it in the large number of ovirt related posts). Has anyone done this or could someone point me in the right direction to docs on how to do this? Trying to even get direct access isn't getting me far (/usr/bin/novnc_server --vnc localhost:5900 -> click URL -> enter password, hangs on "Starting VNC handshake"). Thanks! PS - while testing, iptables and selinux are disabled... I'm trying to get a most basic setup working -- Digimer Papers and Projects: https://alteeve.ca/w/ What if the cure for cancer is trapped in the mind of a person without access to education?

2 1

[libvirt-users] Libvirt from source problem
by Muhammad Asif 21 Oct '15

21 Oct '15

Hi Geeks, When I install libvirt from source, there is no libvirtd. how can i start or restart libvirt deamon. Moreover it give some permission error. What issue can be. Thanks

3 2

[libvirt-users] Libvirt-LXC, container get stuck if it prints too much logs in the console
by Bertrand NOEL 21 Oct '15

21 Oct '15

Hi, I am creating containers using libvirt-lxc. My problem is that a process at bootup gets stuck at some point. My usecase is Puppet, but I can reproduce with a loop in bash. I noticed it is stuck until I connect to the console of the container (virsh -c lxc:/// console id). Then after disconnecting the console, waiting some times, the process gets stuck again. The more the process outputs to the console, the quicker it becomes stuck. To test that behavior, at bootup (in rc.local), in a loop I curl some server that I monitor. In my server logs I can see requests stops after some time. (I tweaked rc.local systemd service to output to console) Host and container are CentOS 7.1 Libvirt v1.2.8 Below is my libvirt.xml <domain type="lxc"> <uuid>9fbe3d43-e1b4-4244-8dab-b42995ab7403</uuid> <name>instance-000002f6</name> <memory>4194304</memory> <vcpu>2</vcpu> <os> <type>exe</type> <cmdline>console=tty0 console=ttyS0</cmdline> <init>/sbin/init</init> </os> <clock offset="utc"/> <devices> <filesystem type="mount"> <source dir="/root/rootfs"/> <target dir="/"/> </filesystem> <interface type="bridge"> <mac address="02:16:3E:00:F6:0E"/> <source bridge="virbr0"/> </interface> <console type="pty"/> </devices> </domain>

1 0

[libvirt-users] "failed to connect to the hypervisor"
by Ken D'Ambrosio 21 Oct '15

21 Oct '15

Hi, all. Ubuntu host, attempting to get virsh working with VirtualBox... and failing. Here's what happens: root@foobox:~# virsh -c vbox:///session list error: failed to connect to the hypervisor error: internal error: unable to initialize VirtualBox driver API ------------------------------------------------------ Now, most of the docs I read about "failed to connect to the hypervisor" have other errors that happen *before* the "failed to connect" one, that are the root cause (e.g., unable to open socket, or bad cert, or whatever). But that's not the case, here. And what puzzles me more is that -- while I don't know what the output *should* look like -- to my eye, strace seems to say that we're getting information back: 1940 read(5, "Node 1 MemTotal: 33011680 kB\nNode 1 MemFree: 32565612 kB\nNode 1 MemUsed: 446068 kB\nNode 1 Active: 57948 kB\nNode 1 Inactive: 77036 kB\nNode 1 Active(anon): 27688 kB\nNode 1 Inactive(anon): 84 kB\nNode 1 Active(file): 30260 kB\nNode 1 Inactive(file): 76952 kB\nNode 1 Unevictable: 52 kB\nNode 1 [...] ------------------------------------------------------ In my googling, I see mention of per-VM XML files that... do something. But what they do, and how they're generated, doesn't seem clear. (E.g., XML mentioned here: http://libvirt.org/drvvbox.html -- it just sort of sits there, with no context.) Is that something I need? Thanks for any enlightenment you might be able to offer me, -Ken

2 4

[libvirt-users] enable aes for guest
by Parvathi 20 Oct '15

20 Oct '15

Hi, How do I enable aes for guest VM? Host CPU supports aes (cat /proc/cpuinfo has aes in its flags and virsh capabilities also displays aes as a feature). When I add <features> <aes/> </features> to the xml file, I am unable to create the VM. It complains with the below error: virsh create /tmp/tmp.xml error: Failed to create domain from /tmp/tmp.xml error: internal error: unexpected feature aes Best!

2 1

[libvirt-users] selective virsh host permissions
by Jamie Fargen 20 Oct '15

20 Oct '15

As a Systems Administrator, I would like to grant permissions to a certain VM using unix groups. In this example there is a hypervisor with VMs A,B,C,D and there is a group called fortadmins. The solution I am searching forI would just allow fortadmins to use libvirt/virsh commands on VM D. Does libvirt/virsh provide any way to easily accomplish this goal? Regards, Jamie Ian Fargen

3 2

[libvirt-users] virsh blockcopy without undefine
by Nathan Shearer 14 Oct '15

14 Oct '15

I have a script that performs a backup with the following steps: virsh dumpxml --inactive "$DOMAIN" > "$TMP/${DOMAIN}.xml" && \ virsh dumpxml --inactive "$DOMAIN" > "$DESTINATION_XML" && \ virsh undefine "$DOMAIN" && \ virsh blockcopy "$DOMAIN" "$SOURCE_DISK" "$DESTINATION_DISK" --wait --verbose --finish && \ virsh define "$TMP/${DOMAIN}.xml" It's basic and did work for creating basic backups. Is there a way to do this without undefining the domain? I recently had to redefine the VM from a backup XML because this script failed during the block copy step. Why can't I blockcopy a domain that is defined? Why does it need to be undefined first?

2 2

[libvirt-users] virsh can't support VM offline blockcommit
by 陈明 13 Oct '15

13 Oct '15

Hi everyone! I use the libvirt(version: 1.2.2) and QEMU(version: 2.2.1) to test qemu snapshot features: I tried virsh blockcommit when VM offline, the virsh blockcommit failed: the error messase as below: * error: Requested operation is not valid: domain is not running* when I start the VM, the virsh blockcommit work fine! my question is : we need virsh support blockcommit when VM offline。how do？ we test qemu snapshot as follows： we need create snapshot of the cycle，and by virsh blockcommit merge snapshot chain，but whenVM offline， the virsh blockcommit will fail， someone suggest me：when VM offline，may use qemu-img commit to submit snapshot turn，but qemu-img can't modify snapshot chain backing relation，it need we use qemu-img rebase amend snapshot‘s backingfile，* the main question* is when we use qemu-img rebase modify snapshot's backingfile, if someone start the VM, it will may lead to VM filesystem readonly。 we should how to solve the question? thanks!

1 0

[libvirt-users] Creating external snapshots.
by Dominique Ramaekers 12 Oct '15

12 Oct '15

Hi, I'm trying to create an external snapshot for backup purposes. Only I get an error after the command: virsh snapshot-create-as --domain PCVIRT1 PCVIRT1-SN1 --diskspec vda,file=/var/lib/libvirt/images/VIRT1-SN1.qcow2 --disk-only --atomic --quiesce fout: internal error: kan AppArmor profiel 'libvirt-2c9c04d7-f86d-df12-da9f-9847ac8b994f' niet vernieuwen (Sorry about the dutch... it means: error: internal error: can not renew AppArmor profile 'libvirt-....') I can't find anything about this error... Anybody got an idea how to resolve this problem? Greetings, Dominique.

2 1

[libvirt-users] virsh can't support VM offline blockcommit
by 陈明 12 Oct '15

12 Oct '15

Hi everyone! I use the libvirt(version: 1.2.2) and QEMU(version: 2.2.1) to test qemu snapshot features: I tried virsh blockcommit when VM offline, the virsh blockcommit failed: the error messase as below: * error: Requested operation is not valid: domain is not running* when I start the VM, the virsh blockcommit work fine! my question is : we need virsh support blockcommit when VM offline。how do？ we test qemu snapshot as follows： we need create snapshot of the cycle，and by virsh blockcommit merge snapshot chain，but whenVM offline， the virsh blockcommit will fail， someone suggest me：when VM offline，may use qemu-img commit to submit snapshot turn，but qemu-img can't modify snapshot chain backing relation，it need we use qemu-img rebase amend snapshot‘s backingfile，* the main question* is when we use qemu-img rebase modify snapshot's backingfile, if someone start the VM, it will may lead to VM filesystem readonly。 we should how to solve the question? thanks!

1 0

[libvirt-users] OPW 2015 application
by Ishani 11 Oct '15

11 Oct '15

Sir, I am Ishani Chugh, a third year undergraduate student of Electronics and Communications Engineering at International Institute of Information Technology, Hyderabad. I am passionate about Open Source Development. I also have some experience working with git, I aspire to contribute to QEMU as Open source Developer under OPW and I am interested in working on the project 'Enhancing libvirt-designer', since I feel, I have the enthusiasm and skills required. I will appreciate any assistance with regards to beginner contribution on this project. Sincerely, Ishani

1 0

[libvirt-users] virsh can't support VM offline blockcommit
by justlibvirt 10 Oct '15

10 Oct '15

Hi everyone! I use the libvirt(version: 1.2.2) and QEMU(version: 2.2.1) to test qemu snapshot features: I tried virsh blockcommit when VM offline, the virsh blockcommit failed: the error messase as below: error: Requested operation is not valid: domain is not running when I start the VM, the virsh blockcommit work fine! my question is : we need virsh support blockcommit when VM offline。how do？ we test qemu snapshot as follows： we need create snapshot of the cycle，and by virsh blockcommit merge snapshot chain，but whenVM offline， the virsh blockcommit will fail， someone suggest me：when VM offline，may use qemu-img commit to submit snapshot turn，but qemu-img can't modify snapshot chain backing relation，it need we use qemu-img rebase amend snapshot‘s backingfile， the main question is when we use qemu-img rebase modify snapshot's backingfile, if someone start the VM, it will may lead to VM filesystem readonly。 we shoule solve the question? thanks!

1 0

[libvirt-users] qemu:///session and network-mounted home directories
by Spencer Baugh 10 Oct '15

10 Oct '15

Dear libvirt-users, Suppose my home directory is network mounted on to several different machines. And suppose I defined some VMs in qemu:///session. The configuration files and disk images for qemu:///session are stored in my home directory. Now suppose I log in to some machine and use virsh. The qemu:///session VMs that are marked autostart are started. Now if I log into another machine and use virsh, it will presumably try to start my autostart VMs again, since the same configuration is present in both homedirs. Is there anything libvirt does that might prevent it from starting the same VM multiple times on different machines? Or might allow machine-specific autostart configuration? Maybe something could be done with /etc/machine-id! In terms of qemu:///system, this is like having /var/lib/libvirt/images/ live on a network-mounted filesystem (fairly normal I assume) and also having /etc/libvirt/qemu live on a network-mounted filesystem (not as normal). Not sure how you would deal with the issue there either but I can see why you might want to have your same VM configuration available on all hosts, so it would be cool to figure out that problem as well. The answer may be the VM configuration is host-specific, so sharing VM configuration between hosts is not a good idea. In that case, any other recommendations on how to deal with the networked homedir case? Thanks!

2 3

[libvirt-users] Setting a VM's vnetX qlen in the XML file
by Digimer 09 Oct '15

09 Oct '15

Hi all, I have a two-node HA RHEL 6 setup hosting VMs. These use static/standard bridges and all has worked wonderfully. Now though, I need to adjust the qlen value of the vnetX devices for a given guest VM, which I can do with: ip link set dev vnet0 qlen 6000 The problem with this approach is two-fold though; 1. I need to determine which vnetX is connected to a VM. 2. If the VM live-migrates, the qlen of the vnetX on the new host needs to be updated. This causes an issue where network traffic is impacted until the command is (re)ran. So I am hoping there is a way I can define the qlen right in the VM's XML definition file so that it is set when the vnetX device is created. Is this possible? Thanks! -- Digimer Papers and Projects: https://alteeve.ca/w/ What if the cure for cancer is trapped in the mind of a person without access to education?

1 0

[libvirt-users] ceph vs gluster as libvirt backend
by Patrick F. Marques 09 Oct '15

09 Oct '15

Hi Hi, I looking for some info about using ceph or gluster as VMs disk backend, anyone has some experience on this? I never used any of them. If someone could share some issues or strengths on integration or even some performance details/stories, I appreciate. Regards, Patrick Marques

1 0

[libvirt-users] autpfs + nfs stuck on stat() inside libvirt lxc 1.2.18
by mxs kolo 08 Oct '15

08 Oct '15

Hi all. Has someone success story with using autofs + nfs inside libvirt LXC container ? In my case nfs client and server in lxc work just fine, but with autofs hang on system call state(). I use CE7_64 on nodes and inside container with libvirt 1.2.18 May be I must set some of capabilities in <feature> (http://man7.org/linux/man-pages/man7/capabilities.7.html) ? I try only CAP_SYS_ADMIN and it's no solve problem. b.r. Maxim Kozin

1 2

[libvirt-users] Libvirt fails to build with sasl
by Jonny Schlag 08 Oct '15

08 Oct '15

Description of problem: I try to build a libvirt package for the IPFire project (http://www.ipfire.org/) Without sasl libvirt build fine but with sasl libvirt fails to build with the following message: CC rpc/libvirt_net_rpc_la-virnetsaslcontext.lo In file included from rpc/virnetsaslcontext.h:25, from rpc/virnetsaslcontext.c:26: /usr/include/sasl/sasl.h:230: error: expected declaration specifiers or '...' before 'size_t' Makefile:7023: recipe for target 'rpc/libvirt_net_rpc_la-virnetsaslcontext.lo' failed make[3]: *** [rpc/libvirt_net_rpc_la-virnetsaslcontext.lo] Error 1 make[3]: Leaving directory '/libvirt-1.2.20/src' Makefile:5119: recipe for target 'all' failed make[2]: *** [all] Error 2 make[2]: Leaving directory '/libvirt-1.2.20/src' Makefile:1999: recipe for target 'all-recursive' failed make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory '/libvirt-1.2.20' Makefile:1896: recipe for target 'all' failed make: *** [all] Error 2 Version-Release number of selected component (if applicable): 1.2.20 How reproducible: Build libvirt with: 1. ./configure --prefix=/usr --localstatedir=/var --sysconfdir=/etc --with-openssl --with-sasl --without-uml --without-vbox --without-lxc --without-esx --without-vmware --without-openvz --without-firewalld --without-storage-scsi --without-storage-mpath --without-avahi --with-test-suite -without-dbus --without-storage-lvm --without-pm-utils 2. make 3. the error occurs Additional info: Cyrus sasl (2.1.26) is build with ./configure --prefix=/usr --sysconfdir=/etc \ --with-dbpath=/var/lib/sasl/sasldb2 \ --with-saslauthd=/var/run/saslauthd \ --with-des=no --with-rc4=no I attached the whole terminal output. Sincerely Jonatan Schlag PS.: Sorry for my bad english, i hope you could understand me.

2 2