February 2011 - Users - libvirt List Archives

Re: [libvirt-users] libvirt unavailable while a VM is in migration?]
by Igor Serebryany 24 Mar '11

24 Mar '11

On Mon, Dec 20, 2010 at 02:14:42PM +0100, Jiri Denemark wrote: > Ouch, I wonder if that could be the reason... Just to make sure that my strange compiling/packages aren't causing the problem, I've set up two servers running fedora 14 with the official libvirt packages. i experience the same problem -- while a vm is in migration, nothing else works. my traces reveal libvirt getting stuck at the same exact spot as on my debian machines. at this point, i'm pretty sure that this is a bug in libvirt. Has libvirt ever had the ability to perform other operations while a qemu migration is in progress? If so than this is a regression, and a pretty serious one. For me, this makes live-migrations totally unusable, and since I cannot run my project without live-migrations and I cannot use any outside tools to do live-migrations when I'm using libvirt this makes libvirt totally unusable. i am desperate to get this bug fixed, and i don't think i have the resources to try to patch it myself. is there anything i can do to help you guys? more debugging info? foot massage? --Igor

2 3

[libvirt-users] Using dlls for Windows provided in http://libvirt.org/sources/win32_experimental/Libvirt-0.8.7-2.exe
by Schnizer, Monika 24 Feb '11

24 Feb '11

Hello libvirt-users, could anybody tell me, whether we could use dlls provided in the experimental package for windows, and distribute it with our application. We would do the following: a) Install the executable from http://libvirt.org/sources/win32_experimental/Libvirt-0.8.7-2.exe on our production machine. b) Add the dlls to the setup of our application c) Install the dlls on the customer's system in some location named /<product root>/lib. Questions: - do we violate the LGPL by doing so? - does anybody have better suggestions on how to provide the functionality to the customer? - who could tell me how to contact the copyright owners, in order to directly discuss this issue with them. Let me thank in advance for any any information. With best regards, -- Monika Schnizer Software Development FTS TSP x86 E SW4 FUJITSU Fujitsu TECHNOLOGY SOLUTIONS GmbH Domagk-Str.28, 80807 Munich, Germany Tel: +49 (89) 3222 2287 Fax: +49 (89) 3222 329 2287 Email: Monika.Schnizer(a)ts.fujitsu.com Web: http://ts.fujitsu.com<http://ts.fujitsu.com/> Company Details: de.ts.fujitsu.com/imprint.html

1 0

[libvirt-users] libvirt cgroup
by Zvi Dubitzky 23 Feb '11

23 Feb '11

1. I am using libvirt 0.8.3 on fedora14 . Acoording to http://berrange.com/posts/2009/12/03/using-cgroups-with-libvirt-and-lxckvm-… inorder to have libvirt memory cgroup under a pre specified group one should add CGROUP_DAEMON="memory:/virt" to /etc/sysconfig/libvirtd and restart cgconfig service and the libvirtd service . That did not help . any idea ? 2. Also : specifying a limit to the memory subsystem (/dev/cgroup/memory/libvirt/qemu/memory.limit_in_bytes) by echoing a value , does not propagate to a VM generated later under qemu which is actually generated with a huge default value for memory.limit_in_bytes . Suppose there was such a limit . Should it decline a generation of a VM that requests a higher memory size in its XML file or the limit prevents the VM from consuming more that the specified cgroup limit when consuming a higher value thanks Zvi Dubitzky Email:dubi@il.ibm.com

1 0

[libvirt-users] Using qemu-kvm divre snapshot option
by Alessandro Italiano 23 Feb '11

23 Feb '11

Hi, looking at the qemu-kvm man page, it is possible to enable snapshot for a given drive. This is what we need because we would like to start a domain with two drives, one on a file with snapshot=on and an other drive on LVM with snapshot=off. Unfortunately it seems that if you use libvirt the same option is not available when you define the domain with a XML file. I mean, I didn't find out a way to support the above use case. So, does libvirt support the drive snapshot option as qemu-kvm does ? [root@dom0-103-13-26-b libvirt-0.8.8]# cat /etc/issue Scientific Linux sl54_x86_64 [root@dom0-103-13-26-b libvirt-0.8.8]# rpm -qa | grep libvirt libvirt-0.6.3-20.1.el5_4.x86_64 libvirt-python-0.6.3-20.1.el5_4.x86_64 [root@dom0-103-13-26-b libvirt-0.8.8]# rpm -qa | grep qemu qemu-common-0.10.5-1.el5.2.x86_64 kvm-qemu-img-83-164.el5_5.21.x86_64 qemu-system-x86-0.10.5-1.el5.2.x86_64 Below you can find a copy of the XML where we would like to add somewhere the option snapshot <domain type='kvm'> <name>centos5_b</name> <uuid>a973d1d4-7af6-3f41-a3c1-bea958aba98b</uuid> <memory>524288</memory> <currentMemory>524288</currentMemory> <vcpu>1</vcpu> <os> <type arch='x86_64' machine='rhel5.4.0'>hvm</type> <boot dev='hd'/> </os> <features> <acpi/> <apic/> <pae/> </features> <clock offset='utc'/> <on_poweroff>destroy</on_poweroff> <on_reboot>restart</on_reboot> <on_crash>restart</on_crash> <devices> <emulator>/usr/libexec/qemu-kvm</emulator> <disk type='file' device='disk'> <driver name='qemu' cache='none'/> <source file='/home/davide/WNoDeS/disks/centos5_b.raw'/> <target dev='vda' bus='virtio'> </disk> <disk type=block device='disk'> <source file='/dev/vg0/lv_slash'/> <target dev='vda' bus='virtio'> </disk> <disk type='file' device='cdrom'> <target dev='hdc' bus='ide'/> <readonly/> </disk> <interface type='bridge'> <mac address='00:16:36:36:70:3e'/> <source bridge='br0'/> <model type='virtio'/> </interface> <serial type='pty'> <target port='0'/> </serial> <console type='pty'> <target port='0'/> </console> <input type='mouse' bus='ps2'/> <graphics type='vnc' port='-1' autoport='yes' keymap='it'/> </devices> </domain> Thanks in advance Alessanndro ********************************************** Alessandro Italiano INFN-CNAF,Tier1 Farming Group Viale Berti Pichat 6/2, 40127 Bologna, Italy Tel: +39 051 6092751 alessandro.italiano(a)cnaf.infn.it www.cnaf.infn.it Skype#alessandroitaliano **********************************************

1 0

[libvirt-users] [PATCH 1/2] libvirt/qemu : allow persistent modification of disks via A(De)ttachDeviceFlags
by KAMEZAWA Hiroyuki 23 Feb '11

23 Feb '11

>From 030135224dd6563af0fb8615dc6a4b8e6084410d Mon Sep 17 00:00:00 2001 From: KAMEZAWA Hiroyuki <kamezawa(a)bluextal.(none)> Date: Wed, 23 Feb 2011 15:25:26 +0900 Subject: [PATCH 1/2] libvirt/qemu : support attach/detach-disk --persistent Now, only Xen supports 'virsh attach/detach-disk XXXX --persistent', modifying inactive domain definition via virsh. This patch adds a support for qemu. With this patch, virsh attach/detach-disk --persistent works well with qemu. Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu(a)jp.fujitsu.com> --- src/qemu/qemu_driver.c | 170 ++++++++++++++++++++++++++++++++++++++++++++++-- 1 files changed, 163 insertions(+), 7 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 0f25a2a..703f86a 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -4082,16 +4082,174 @@ cleanup: return ret; } +static int qemuDomainFindDiskByName(virDomainDefPtr vmdef, const char *name) +{ + virDomainDiskDefPtr vdisk; + int i; + + for (i = 0; i < vmdef->ndisks; i++) { + vdisk = vmdef->disks[i]; + if (!strcmp(vdisk->dst, name)) + return i; + } + return -1; +} +/* + * Attach a device given by XML, the change will be persistent + * and domain XML definition file is updated. + */ +static int qemuDomainAttachDevicePersistent(virDomainDefPtr vmdef, + virDomainDeviceDefPtr newdev) +{ + virDomainDiskDefPtr disk; + + /* At first, check device confliction */ + switch(newdev->type) { + case VIR_DOMAIN_DEVICE_DISK: + disk = newdev->data.disk; + if (qemuDomainFindDiskByName(vmdef, disk->dst) >= 0) { + qemuReportError(VIR_ERR_INVALID_ARG, + _("target %s already exists."), disk->dst); + return -1; + } + + if (virDomainDiskInsert(vmdef, disk)) { + virReportOOMError(); + return -1; + } + if (disk->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) { + /* + * Address/Drive information is NULL. If virtio, PCI address + * shoule be fixed. Other devices as IDE, SCSI...will get ID + * automatically + */ + qemuDomainAssignPCIAddresses(vmdef); + + } + newdev->data.disk = NULL; + break; + default: + qemuReportError(VIR_ERR_INVALID_ARG, "%s", + _("Sorry, the device is not suppored for now")); + return -1; + } + + return 0; +} + +static int qemuDomainDetachDevicePersistent(virDomainDefPtr vmdef, + virDomainDeviceDefPtr device) +{ + int x; + virDomainDiskDefPtr disk; + + switch(device->type) { + case VIR_DOMAIN_DEVICE_DISK: + disk = device->data.disk; + x = qemuDomainFindDiskByName(vmdef, disk->dst); + if (x < 0) { + qemuReportError(VIR_ERR_INVALID_ARG, + _("target %s doesn't exist."), disk->dst); + return -1; + } + virDomainDiskRemove(vmdef, x); + break; + default: + qemuReportError(VIR_ERR_INVALID_ARG, "%s", + _("Sorry, the device is not suppored for now")); + return -1; + } + return 0; +} + +static int qemuDomainModifyDevicePersistent(virDomainPtr dom, + const char *xml, + unsigned int attach) +{ + struct qemud_driver *driver; + virDomainDeviceDefPtr device; + virDomainDefPtr vmdef; + virDomainObjPtr vm; + int ret = -1; + + if (!dom || !dom->conn || !dom->name || !xml) { + qemuReportError(VIR_ERR_INVALID_ARG, + _("internal error : %s"), __FUNCTION__); + return -1; + } + + if (dom->conn->flags & VIR_CONNECT_RO) + return -1; + + driver = dom->conn->privateData; + qemuDriverLock(driver); + vm = virDomainFindByUUID(&driver->domains, dom->uuid); + if (!vm) { + qemuReportError(VIR_ERR_NO_DOMAIN, _("cannot find domain '%s'"), + dom->name); + goto unlock_out; + } + + if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) { + /* + * For now, just allow updating inactive domains. Further development + * will allow updating both active domain and its config file at + * the same time. + */ + qemuReportError(VIR_ERR_INVALID_ARG, + _("cannot update alive domain : %s"), __FUNCTION__); + goto endjob; + } + vmdef = virDomainObjGetPersistentDef(driver->caps, vm); + + if (!vmdef) + goto endjob; + + device = virDomainDeviceDefParse(driver->caps, + vmdef, xml, VIR_DOMAIN_XML_INACTIVE); + if (!device) + goto endjob; + + if (attach) { + ret = qemuDomainAttachDevicePersistent(vmdef, device); + if (ret < 0) + goto out; + } else { + ret = qemuDomainDetachDevicePersistent(vmdef, device); + if (ret < 0) + goto out; + } + ret = virDomainSaveConfig(driver->configDir, vmdef); + +out: + virDomainDeviceDefFree(device); +endjob: + if (qemuDomainObjEndJob(vm) == 0) + vm = NULL; + if (vm) + virDomainObjUnlock(vm); + /* Note: insert of newdev is done by copy */ +unlock_out: + qemuDriverUnlock(driver); + return ret; +} + static int qemudDomainAttachDeviceFlags(virDomainPtr dom, const char *xml, unsigned int flags) { if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) { - qemuReportError(VIR_ERR_OPERATION_INVALID, - "%s", _("cannot modify the persistent configuration of a domain")); - return -1; + /* + * Because we can't update the live guest and XML + * in atomic, limiting modification as only-acrive and + * only-inactive. Need some idea to update both at the same time. + */ + return qemuDomainModifyDevicePersistent(dom, xml, 1); } - return qemudDomainAttachDevice(dom, xml); + if (flags & VIR_DOMAIN_DEVICE_MODIFY_LIVE) + return qemudDomainAttachDevice(dom, xml); + + return -1; } @@ -4304,9 +4462,7 @@ static int qemudDomainDetachDeviceFlags(virDomainPtr dom, const char *xml, unsigned int flags) { if (flags & VIR_DOMAIN_DEVICE_MODIFY_CONFIG) { - qemuReportError(VIR_ERR_OPERATION_INVALID, - "%s", _("cannot modify the persistent configuration of a domain")); - return -1; + return qemuDomainModifyDevicePersistent(dom, xml, 0); } return qemudDomainDetachDevice(dom, xml); -- 1.7.1

1 0

[libvirt-users] Vista guest doesn't start after libvirt upgrade
by Matthias Meyer 22 Feb '11

22 Feb '11

Hi, I've updated my Debian lenny to Debian squeeze last week. So I have now: libvirt 0.8.3-5 qemu-kvm 0.12.5+dfsg-3 virt-manager 0.8.4-8 Today I try to start my Vista quest again. I am surprised because it won't starts. In "/var/log/libvirt/qemu/devel-vista.log" stands "Read failed". The Console stays as a black screen after a short SeaBIOS message. I tried system recovery too but Vista system recovery identified the Vista installation without any errors. Nearly the same occurs with my Windows7 guest. The logfile seems fine but the console present me: Starting SeaBIOS (version...) Booting from Hard Disk... Boot failed: not a bootable disk I tried system recovery too and indeed, W7 doesn't found a Windows Installation. linux-dev:~# LANG==C ls -lh --block-size=1M /var/lib/libvirt/images/ total 90399 -rwxr-xr-x 1 root root 10000 Feb 12 19:20 devel-CentOS.img -rwxr-xr-x 1 root root 10000 Feb 15 20:17 devel-SUSE.img -rwxr-xr-x 1 root root 10001 Feb 16 22:55 devel-fedora.img -rwxr-xr-x 1 root root 10000 Feb 16 19:23 devel-ubuntu.img -rwxr-xr-x 1 root root 12575 Feb 1 07:56 devel-vista.qcow2 -rwxr-xr-x 1 libvirt-qemu kvm 40961 Feb 21 23:50 devel-vista.qcow2.2del -rwxr-xr-x 1 root root 9330 Jan 23 18:03 devel-w7.qcow2 -rwxr-xr-x 1 root root 10136 Feb 13 20:58 devel-w7.qcow2.2del -rwxr-xr-x 1 root root 10241 Feb 22 00:34 devel-xp.qcow2 the *.2del files are the original images which doesn't start. The responding files without .2del are restored from a backup. WindowsXP as well as different Linux guests does start and run fine. There are any experience with this obscurse behavior? Thanks in advance Matthias -- Don't Panic

2 1

[libvirt-users] schedinfo for kvm
by Zvi Dubitzky 21 Feb '11

21 Feb '11

Is there any schedinfo support for kvm currently ? Zvi Dubitzky Email:dubi@il.ibm.com

2 1

Re: [libvirt-users] LXC libvirt support
by Nirmal Guhan 18 Feb '11

18 Feb '11

2011/2/11 Nirmal Guhan <vavatutu(a)gmail.com>: > On Fri, Feb 11, 2011 at 9:05 PM, Osier Yang <jyang(a)redhat.com> wrote: >> 于 2011年02月12日 11:52, Nirmal Guhan 写道: >>> >>> Hi, >>> >>> Has anyone used libvirt to start/stop/manage LXC (linux containers) ? >>> Is it supported at all? Last I tried, had few issues and the reply was >>> : >> >> Yes, it's supported, though still much work to do, but the basic is >> there. >> >>> >>> /// >>> Note that libvirt lxc driver is a different implementation from lxc tools >>> and doesn't use lxc tools internally. >> >> Indeed, it doesn't invoke lxc tools. >> >> You may not be able to do what >>> >>> you are doing with lxc tools. Especially, the driver doesn't take aware >>> of files and settings inside a container at all. You have to prepare them >>> by yourself in advance >> >> libvirt.org has the example, e.g. the busybox example, I guess it could >> explain "the driver doesn't take care of files and setting inside a >> container at all, you have to prepare them by yourself in advance." >> >> Regards >> Osier > > Thanks. Will give a try. > ~nirmal > Seeing some strange behavior running fedora 12 for both host and lxc container. I do a define, start and console : [root@fedora12new lxc-libvirt-scripts]# virsh -c lxc:// console testf12new Connected to domain testf12new Escape character is ^] --> I press enter here and virsh exits. I don't get login prompt always. Sometimes I do and when I try to login it rejects my password. Apparently, the same rootfs works fine with lxc tools and I can login. Snippet from etc/event.d/tty1 : respawn exec /sbin/mingetty pts/0 I use "exec /sbin/mingetty tty1" for lxc tools and that works fine. I had to try pts/0 for libvirt since tty1 does not even get me the occasional login prompts I get with pts. Also even after virsh shutdown, I see that the container processes do not exit. Is there something else I should do? ~nirmal

3 2

[libvirt-users] altering virtual network driver iptables behavior
by Ryan Sumner 18 Feb '11

18 Feb '11

I have the need to modify the behavior of the virtual network driver's behavior and how it deals with routed networks. I'm running libvirt-0.8.3-2.fc14. According to http://libvirt.org/firewall.html, the following is automatically added to the FORWARD chain of iptables when a network type of "routed" is started up: "Allow inbound, but only to our expected subnet. Allow outbound, but only from our expected subnet. Allow traffic between guests. Deny all other inbound. Deny all other outbound. " The part of this that I need to adjust is the fact that only IPs on my subnet will be allowed in and out. I have IP addresses assigned to my guests that have static routes configured on the hypervisor to route to the local bridged interface. I have to do this due to the way the surrounding routers and switches on the network are configured to handle public IP addresses and MAC address filtering. Here is an example of my config where the public IP address on the guest machine is 1.1.2.2: My network to do the routing mode: <network> <name>local</name> <forward dev='eth0' mode='route'/> <bridge name='virbr_local' stp='on' delay='0' /> <ip address='192.168.122.1' netmask='255.255.255.0'/> </network> The network portion of the domain: <interface type='network'> <mac address='xx:xx:xx:cc:xx:xx'/> <source network='local'/> <target dev='vnet0'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x02' function='0x0'/> </interface> And finally the static route configured on the hypervisor to get the routed traffic from eth0 on the hypervisor down to the virbr_local device: # ip route add 1.1.2.2 dev virbr_local This will setup the iptables filters just as the documentation defines, but the problem is that all traffic from the guest will get REJECTED by iptables due to the source and destination IP not falling within 192.168.122.0/24. I've tried adding a custom filter into the network filter driver, but haven't had much luck. Here are some of the things that I've tried. The custom network filter. Notice that I'm using tcp, udp, and icmp specifically. I'm doing this so it will force inclusion into the iptables filtering rules rather than ebtables. <filter name='my-static-ip' chain='root'> <rule action='accept' direction='out' priority='500'> <tcp srcipaddr='$MYIP'/> </rule> <rule action='accept' direction='out' priority='500'> <udp srcipaddr='$MYIP'/> </rule> <rule action='accept' direction='out' priority='500'> <icmp srcipaddr='$MYIP'/> </rule> <rule action='accept' direction='in' priority='500'> <tcp dstipaddr='$MYIP'/> </rule> <rule action='accept' direction='in' priority='500'> <udp dstipaddr='$MYIP'/> </rule> <rule action='accept' direction='in' priority='500'> <icmp dstipaddr='$MYIP'/> </rule> </filter> And the modifications made to the domain's network interface definition: <interface type='network'> <mac address='xx:xx:xx:xx'/> <source network='local'/> <target dev='vnet0'/> <filterref filter='my-static-ip'> <parameter name='MYIP' value=1.1.2.2'/> </filterref> <address type='pci' domain='0x0000' bus='0x00' slot='0x02' function='0x0'/> </interface> With the exception of simply manually adding iptables rules in place after I start the network using virsh, does anyone know how to accomplish what I'm trying to do? Ryan Sumner

1 0

[libvirt-users] schedinfo
by Zvi Dubitzky 16 Feb '11

16 Feb '11

libvirt schedinfo : ------------------ What is the meaning of the 'cap' parameter in the schedinfo virsh command (what are the units and range of values ) and what is the 'weigth' parameter ? Zvi Dubitzky Email:dubi@il.ibm.com

1 0

[libvirt-users] Clarification about virsh migration options
by Berend Dekens 16 Feb '11

16 Feb '11

I am trying to work out what all the options are for migrating a KVM machine to another KVM machine, without using shared storage. The documentation is not quite verbose and not intuitive, so I'm hoping someone can explain this to me. The man pages show this syntax: migrate optional --live --p2p --direct --tunnelled --persistent --undefinesource --suspend --copy-storage-all --copy-storage-inc domain-id desturi migrateuri dname The 'live' and 'suspend' options are clear. The 'undefinesource' option is straightforward as well. But what does 'persistent' mean? I mean, when transferring a VM to a destination, it will be available on the destination when migration completes, so what does 'persistent' mean in this context? The p2p, direct and tunneled options are not. When migrating, I assume it is possible to let the underlying virtualisation framework handle the migration - so I assume 'direct' means Qemu-KVM gets to migrate the VM and 'tunneled' means the RPC mechanism of libvirt migrates the machine. But what does 'p2p' mean? Normally, peer-to-peer implies direct communication, but since there is a 'direct' mode, I'm clueless what this option does. On a side note: is it possible to use tunneled mode to transfer VMs from for example Xen to KVM? Then I have some questions about the non-shared storage migration. I really like this option as most of my virtualized servers are run on one or two physical systems without shared VM storage. Migrating those VMs without downtime would be awesome. If I migrate the VMs, I assume the storage of the VM is places in the libvirt default storage location. But what is the difference between 'copy-storage-all' and 'copy-storage-inc'? Incremental hints that it would save incremental changes without actually transferring the storage between hosts. But how is that possible if the storage was not shared in the first place? Lots of questions but I hope someone who has done this before can answer some of these things. Regards, Berend Dekens

3 3

[libvirt-users] CfP 6th Workshop on Virtualization in High-Performance Cloud Computing (VHPC'11)
by VHPC2011 16 Feb '11

16 Feb '11

Apologies if you received multiple copies of this message. ================================================================= CALL FOR PAPERS 6th Workshop on Virtualization in High-Performance Cloud Computing VHPC'11 as part of Euro-Par 2011, Bordeaux, France ================================================================= Date: August 30, 2011 Euro-Par 2011: http://europar2011.bordeaux.inria.fr/ Workshop URL: http://vhpc.org SUBMISSION DEADLINE: Abstracts: May 2, 2011 Full Paper: June 13, 2011 Scope: Virtualization has become a common abstraction layer in modern data centers, enabling resource owners to manage complex infrastructure independently of their applications. Conjointly virtualization is becoming a driving technology for a manifold of industry grade IT services. The cloud concept includes the notion of a separation between resource owners and users, adding services such as hosted application frameworks and queuing. Utilizing the same infrastructure, clouds carry significant potential for use in high-performance scientific computing. The ability of clouds to provide for requests and releases of vast computing resource dynamically and close to the marginal cost of providing the services is unprecedented in the history of scientific and commercial computing. Distributed computing concepts that leverage federated resource access are popular within the grid community, but have not seen previously desired deployed levels so far. Also, many of the scientific datacenters have not adopted virtualization or cloud concepts yet. This workshop aims to bring together industrial providers with the scientific community in order to foster discussion, collaboration and mutual exchange of knowledge and experience. The workshop will be one day in length, composed of 20 min paper presentations, each followed by 10 min discussion sections. Presentations may be accompanied by interactive demonstrations. It concludes with a 30 min panel discussion by presenters. TOPICS Topics include, but are not limited to, the following subjects: - Virtualization in cloud, cluster and grid environments - VM-based cloud performance modeling - Workload characterizations for VM-based environments - Software as a Service (SaaS) - Cloud reliability, fault-tolerance, and security - Cloud, cluster and grid filesystems - QoS and and service levels - Cross-layer VM optimizations - Virtualized I/O and storage - Virtualization and HPC architectures including NUMA - System and process/bytecode VM convergence - Paravirtualized driver development - Research and education use cases - VM cloud, cluster distribution algorithms - MPI on virtual machines and clouds - Cloud frameworks and API sets - Checkpointing of large compute jobs - Cloud load balancing - Accelerator virtualization - Instrumentation interfaces and languages - Hardware support for virtualization - High-performance network virtualization - Auto-tuning of VMM and VM parameters - High-speed interconnects - Hypervisor extensions and tools for cluster and grid computing - VMMs/Hypervisors - Cloud use cases including optimizations - Performance modeling - Fault tolerant VM environments - VMM performance tuning on various load types - Cloud provisioning - Virtual machine monitor platforms - Pass-through VM device access - Management, deployment of VM-based environments PAPER SUBMISSION Papers submitted to the workshop will be reviewed by at least two members of the program committee and external reviewers. Submissions should include abstract, key words, the e-mail address of the corresponding author, and must not exceed 10 pages, including tables and figures at a main font size no smaller than 11 point. Submission of a paper should be regarded as a commitment that, should the paper be accepted, at least one of the authors will register and attend the conference to present the work. Accepted papers will be published in the Springer LNCS series - the format must be according to the Springer LNCS Style. Initial submissions are in PDF, accepted papers will be requested to provided source files. Format Guidelines: http://www.springer.de/comp/lncs/authors.html Submission Link: http://edas.info/newPaper.php?c=10155 CHAIR Michael Alexander (chair), IBM, Austria Gianluigi Zanetti (co-chair), CRS4, Italy PROGRAM COMMITTEE Paolo Anedda, CRS4, Italy Volker Buege, University of Karlsruhe, Germany Giovanni Busonera, CRS4, Italy Roberto Canonico, University of Napoli, Italy Tommaso Cucinotta, Scuola Superiore Sant'Anna, Italy William Gardner, University of Guelph, Canada Werner Fischer, Thomas-Krenn AG, Germany Wolfgang Gentzsch, Max Planck Gesellschaft, Germany Marcus Hardt, Forschungszentrum Karlsruhe, Germany Sverre Jarp, CERN, Switzerland Shantenu Sjha, Louisiana State University, USA Xuxian Jiang, NC State, USA Kenji Kaneda, Google, USA Simone Leo, CRS4, Italy Ignancio Llorente, Universidad Complutense de Madrid, Spain, Naoya Maruyama, Tokyo Institute of Technology, Japan Jean-Marc Menaud, Ecole des Mines de Nantes, France Anastassios Nanos, National Technical University of Athens, Greece Jose Renato Santos, HP Labs, USA Deepak Singh, Amazon Webservices, USA Boria Sotomayor, University of Chicago, USA Yoshio Turner, HP Labs, USA Kurt Tutschku, University of Vienna, Austria Lizhe Wang, Indiana University, USA Chao-Tung Yang, Tunghai University, China DURATION: Workshop Duration is one day. GENERAL INFORMATION The workshop will be held as part of Euro-Par 2011, organized by INRIA, CNRS and the University of Bordeaux I, II, France. Euro-Par 2011: http://europar2011.bordeaux.inria.fr/

1 0

[libvirt-users] Use libvirt-python to attach cdrom
by seven 16 Feb '11

16 Feb '11

Hi all, I want to attach cdrom by libvirt.py .But can't find a func in libvirt.py can finish this work. Who can help me ? Regard -- Seven Ling E-mail: jiejie.ling(a)qq.com

2 3

[libvirt-users] monitiring cpu usage via cgroup
by Zvi Dubitzky 16 Feb '11

16 Feb '11

thanks Kame, 1) the policy of my process(test) is indeed SCHED_NORMAL . When I use sched_setscheduler() to set the pid sched policy to SCHED_FIFO then the cpu usage via cgroup rt_runtime_us and rt_period_us is affected . Still the ratio I specified is 1:10 but I see via top that my process ,that gets 1 cpu ,shows 20% usage 2) What about memory and cpuset subsystems of cgroup. Do their settings apply also only in the cases of processes with policy = SCHED_FIFO/SCHED_RR ? 3) lastly: What is the meaning of the ‘cap’ parameter in the schedinfo virsh command (what are the units and range of values ) and what is the ‘weigth’ parameter .I did not find any documentation of libvirt about it ? thanks Zvi Dubitzky Email:dubi@il.ibm.com

1 0

[libvirt-users] monitiring cpu usage via cgroup
by Zvi Dubitzky 15 Feb '11

15 Feb '11

Hi I was asking about the fedora 14 kernel if it is good enough for cgroup usage because I am trying to set a cgroup under cpu subsytem ( /dev/cgroup/cpu/group1/ ) that have /cpu.rt_runtime_us of 100000 while cpu.rt_period_us has a value of 1000000 i.e a ratio of 1/10 . still when I run a task (endless loop) in that group (cgexec -g cpu,cpuset:group1 ./test) it gets all the cpu core time that is assigned to it ( i watch via top utility) , so it seems that the quota set via the group does not take effect although I restart the cgconfig service . I also verified libcgroup is installed by: rpm -q libcgroup Am I missing something or top is not the right utility to watch the cpu usage in this case ? thanks Zvi Dubitzky Email:dubi@il.ibm.com

2 1

[libvirt-users] Xen disk device detach fails as non-root [libvirt-0.8.7 and older versions]
by Iain MacDonnell 15 Feb '11

15 Feb '11

Hi All, I find that I am able to attach a disk device do a Xen domain, using virDomainAttachDevice(), running as a non-root user, but I am unable to use virDomainDetachDevice() - it results in an "unknown failure". Using "virsh [attach|detach]-device" exhibits this behviour. $ virsh attach-device domu1 attach.xml Device attached successfully $ virsh detach-device domu1 attach.xml error: Failed to detach device from attach.xml error: Unknown failure $ With some digging, I determined that the problem arises when libvirt tries to translate the device name to a number, using the XenStore API (xenStoreDomainGetDiskID()), which requires use of the "xenstored" UNIX socket, and that socket is only accessible by root. On making that socket accessible to the user (by group), virDomainAttachDevice() starts working, but I'm then unable to list domains, because xenStoreDoListDomains() waits to verify each domain using xenHypervisorHasDomain(), and that requires access to another socket - "/proc/xen/privcmd" My question, before going down the path of trying to hack permissions for these sockets permanently ..... is this how it's supposed to be, or could, perhaps, libvirtd, which runs as root, access these sockets on behalf of the user? It seems it should at least fail more gracefully.... TIA for any pointers.... ~iain

2 1

[libvirt-users] connecting to virtualbox with libvirt
by Gary Scarborough 14 Feb '11

14 Feb '11

I am trying to connect to the virtualbox hypervisor through libvirt. If I am doing it correctly, virsh -c vbox:///session should get me connected. I am on F14 64 bit. I tried this with both vbox 3.2 and 4.02. I keep getting the following error: error: no connection driver available for vbox:///session. What am I missing? Thanks, Gary

2 1

[libvirt-users] Linux version that supports cgroup
by Zvi Dubitzky 13 Feb '11

13 Feb '11

Hi Does Fedora release 14 (Laughlin) having Linux kernel 2.6.35.6-45.fc14.x86_64 support cgroup at the kernel ? thanks Zvi Dubitzky Email:dubi@il.ibm.com

2 1

[libvirt-users] cgroup
by Zvi Dubitzky 13 Feb '11

13 Feb '11

Hi Should libvirtd be in a cgroup hierarchy rooted at /dev/cgroup or it can be rooted also at /cgroup i,e can /etc/cgconfig.conf look like his (without the /dev) : mount { cpuset = /cgroup/cpuset; cpu = /cgroup/cpu; cpuacct = /cgroup/cpuacct; memory = /cgroup/memory; devices = /cgroup/devices; freezer = /cgroup/freezer; net_cls = /cgroup/net_cls; ns = /cgroup/ns; blkio = /cgroup/blkio; } thanks Zvi Dubitzky Email:dubi@il.ibm.com

2 1

[libvirt-users] Using NFS to read image file
by Marcela Castro León 12 Feb '11

12 Feb '11

Hello: When I start the guest with an image on NFS file system, the ubuntu doesn't boot, It's remain on inittrams. I've tried to redefine the guest without <apic> <pae> configuration as I see on a blog, and I've increased the parameters of the mount to rsize=32768 and wsize=32768, but it still doesn't work. Can anyone give any advice to get it? Thank you very much. Marcela

2 2

[libvirt-users] LXC libvirt support
by Nirmal Guhan 12 Feb '11

12 Feb '11

Hi, Has anyone used libvirt to start/stop/manage LXC (linux containers) ? Is it supported at all? Last I tried, had few issues and the reply was : /// Note that libvirt lxc driver is a different implementation from lxc tools and doesn't use lxc tools internally. You may not be able to do what you are doing with lxc tools. Especially, the driver doesn't take aware of files and settings inside a container at all. You have to prepare them by yourself in advance /// Is it still the case? ~nirmal

2 1

[libvirt-users] Start domain with usable QMP connection?
by Bryan D. Payne 11 Feb '11

11 Feb '11

Is there any way to get virsh / libvirt to create a domain such that I can later connect to it via QMP (without hacking the code)? The command line that libvirt creates for my current setup looks like this: /usr/local/bin/qemu-system-x86_64 -S -M pc-0.12 -enable-kvm -m 512 -smp 1,sockets=1,cores=1,threads=1 -name winxpsp2 -uuid ab244939-c804-1aff-68b6-9324f174c3c8 -nodefconfig -nodefaults -chardev socket,id=monitor,path=/usr/local/var/lib/libvirt/qemu/winxpsp2.monitor,server,nowait -mon chardev=monitor,mode=control -rtc base=localtime -boot c -drive file=/var/lib/libvirt/images/winxpsp2.img,if=none,id=drive-ide0-0-0,format=raw -device ide-drive,bus=ide.0,unit=0,drive=drive-ide0-0-0,id=ide0-0-0 -drive file=/home/username/iso/en_winxp_pro_with_sp2.iso,if=none,media=cdrom,id=drive-ide0-1-0,readonly=on,format=raw -device ide-drive,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0 -netdev tap,fd=16,id=hostnet0 -device rtl8139,netdev=hostnet0,id=net0,mac=52:54:00:4d:1d:1b,bus=pci.0,addr=0x3 -chardev pty,id=serial0 -device isa-serial,chardev=serial0 -usb -device usb-tablet,id=input0 -vnc 127.0.0.1:0 -k en-us -vga std -device ES1370,id=sound0,bus=pci.0,addr=0x4 -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x5 The important pieces of this command line (I believe) are: -chardev socket,id=monitor,path=/usr/local/var/lib/libvirt/qemu/winxpsp2.monitor,server,nowait -mon chardev=monitor,mode=control However, I'm unable to connect using this path because -- I'm guessing -- libvirt has already connected. As a test, I added some code to src/qemu/qemu_command.c to open a second chardev. My new command line looks like this: /usr/local/bin/qemu-system-x86_64 -S -M pc-0.12 -enable-kvm -m 512 -smp 1,sockets=1,cores=1,threads=1 -name winxpsp2 -uuid ab244939-c804-1aff-68b6-9324f174c3c8 -nodefconfig -nodefaults -chardev socket,id=monitor,path=/usr/local/var/lib/libvirt/qemu/winxpsp2.monitor,server,nowait -mon chardev=monitor,mode=control -chardev socket,id=qmpmonitor,path=/usr/local/var/lib/libvirt/qemu/winxpsp2.qmpmonitor,server,nowait -mon chardev=qmpmonitor,mode=control -rtc base=localtime -boot c -drive file=/var/lib/libvirt/images/winxpsp2.img,if=none,id=drive-ide0-0-0,format=raw -device ide-drive,bus=ide.0,unit=0,drive=drive-ide0-0-0,id=ide0-0-0 -drive file=/home/username/iso/en_winxp_pro_with_sp2.iso,if=none,media=cdrom,id=drive-ide0-1-0,readonly=on,format=raw -device ide-drive,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0 -netdev tap,fd=15,id=hostnet0 -device rtl8139,netdev=hostnet0,id=net0,mac=52:54:00:4d:1d:1b,bus=pci.0,addr=0x3 -chardev pty,id=serial0 -device isa-serial,chardev=serial0 -usb -device usb-tablet,id=input0 -vnc 127.0.0.1:0 -k en-us -vga std -device ES1370,id=sound0,bus=pci.0,addr=0x4 -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x5 Notice that there are now two chardevs: -chardev socket,id=monitor,path=/usr/local/var/lib/libvirt/qemu/winxpsp2.monitor,server,nowait -mon chardev=monitor,mode=control -chardev socket,id=qmpmonitor,path=/usr/local/var/lib/libvirt/qemu/winxpsp2.qmpmonitor,server,nowait -mon chardev=qmpmonitor,mode=control As expected, I am able to connect to this second chardev via QMP. Is there anyway that I can enable such functionality in libvirt without hacking the code? Thanks, bryan

2 3

[libvirt-users] Not able to capture detailed CPU information of the guest machine using Libvirt API.
by arpita k 11 Feb '11

11 Feb '11

Hi , I have two KVM guests in ubuntu host machine.I am using Python binding of Libvirt API to query on the hypervisor and capture the CPU , memory related information of the guest machines. I need to capture the detail information regarding CPU like : cpu_aidle, cpu_idle, cpu_speed, cpu_wio and memory like :mem_cached,mem_buffers,mem_free etc. of the guest machines. How could I get these information? Please help me. Thanks in Advance!! Arpita

3 2

Re: [libvirt-users] [libvirt] configuring network interface
by Marcela Castro León 11 Feb '11

11 Feb '11

Hello Laine: I've revised the configuration and followed all your instructions. I'll try to give you a better explain of my problem. a) I need a bridged (3) configuration, because I have to connect the guests from each others. (via ssh and execute mpi). I've three hosts, each of one has at least a guest. I'll only explain about the configuration of one of them, because the problem is only related a one configuration. hostname=rionegro ip=192.168.1.4 guest-name=rnompi1 b) I've configured host bridge following step by step ther instruction on http://wiki.libvirt.org/page/Networking this is the /etc/network/config # The loopback network interface auto lo iface lo inet loopback auto br0 iface br0 inet static address 192.168.1.4 netmask 255.255.255.0 network 192.168.1.0 broadcast 192.168.1.255 gateway 192.168.1.10 bridge_ports eth0 bridge_stp on bridge_fd 0 bridge_maxwait 0 c) this is the xml guest configuration related to the network. <interface type='bridge'> <source bridge='br0'/> </interface> d) But the guest doesn't get any IP automatically, and create an interface with a variable name (eth4, eth5,etch6..)*. I don't know how to achieve the guest don't behave like this and create always its network interface *with a fixed name, for example, eth0). Why I ask that? if the guest created an interface with a fixed name, I would configure and static ip for the it. I'm attaching a printscreen "rnompi1.jpg" that show what ifconfig ahow on the guest (eth14 and without any IP). It doesn't get their host either. The only way I achieve to solve but "temporally" until reboot is configuring the guest statically using the name of the interface that has just up.In that case every works fine, and the guest can connect to all the other guests on the lan and others can connect to it. But this only works until reboot or reconfigure the guest, because each time the name of interface change, and my configuration I had done of /etc/network/interface didn't work any more because is related to another interface. I hope I've explained better the problem I have. Thank you very much, you've already helped too much to better understand the different cases of network configuration Regards. 2011/2/10 Marcela Castro León <mcastrol(a)gmail.com> > Hello Laine: > Thank you very much. > I've showed only one of the guest, each one have it's own ip address. > Anyway, I'll follow your advise of revise all the configuration after > reading the doc. > Thank you very much again. > Regards. > > 2011/2/8 Laine Stump <laine(a)laine.org> > > (For future questions such as this, you may find you get a better response >> (and lower signal-noise ratio) by asking on the recently -created >> libvirt-users(a)redhat.com list, as libvir-list has a very large volume of >> patches and discussion about the internals of libvirt) >> >> >> On 02/08/2011 12:05 PM, Marcela Castro León wrote: >> >> Hello >> >>> >>> I've a lan of virtual machine configured with bridge. I've a private >> lan of 4 machine, but one of them have 2 lan-card with access to internet. >> >> Evertything works almost fine, except that I've configured on each guest >> the static configuration of the bridge with the ip address, like this: >> >> auto lo >> iface lo inet loopback >> >> auto eth6 >> iface eth6 inet static >> address 192.168.1.41 >> network 192.168.1.0 >> netmask 255.255.255.0 >> gateway 192.168.1.10 >> >> >> >> So are you saying that you're attempting to give each guest the *same* IP >> address? There's no way that can possibly work, even if each is connected to >> a separate virtual network or bridge. Since the host's IP stack can "see" >> all of the guests' network traffic, regardless of the connection mode you >> choose, each must have a unique IP address even if they can't see each >> other. >> >> >> >> >> But each time I, for example, define the guest, it start with a ascending >> number of interface: eth6,eth7, eth8, etc, Now is initializing with eth12. >> >> >> There must be a mixup in the explanation - I'm understanding that you put >> the same config file in each guest, but each guest, when it starts, shows a >> different ethernet interface in its ifconfig. That doesn't seem right, >> because none of the guests know about each other. Or are you saying >> something different? Where are you seeing the "eth6", "eth7", ... "eth12"? >> In the guest XML config maybe? >> >> >> So, it doesn't get the manual configuration I've done on the >> /etc/network/interfaces. >> How can avoid this way of work of the number of ethernet interface the >> guest start? >> >> >> Why are you doing a manual/static definition on the guest at all? It's >> simpler to just leave the guest to automatically configure its eth0, and if >> you really want a particular IP address for a particular guest, handle that >> with a <host> element in the <dhcp> section of the virtual network the guest >> is connected to (or, if you're using a pure bridging setup, then in the >> static hosts config of the dhcp server on that network). >> >> >> >> A way i'm trying to solve is configuring the interface trough the guest >> xml definition, >> >> >> The guest xml interface configuration cannot be used to setup the IP >> address, route, etc config of the guest. It is only used for two things: >> >> 1) configuring what kind of hardware appears on the guest (but not what >> name the guest gives to it!) >> >> 2) configuring how that hardware is connected to the physical network >> (usually either via an existing bridge interface on the host (in order for >> the guest to appear as if its directly connected to the physical network), >> or via a "virtual network" created by libvirt (which is really just a bridge >> that's not directly connected to any physical interface, relying instead on >> the host's IP stack to route packets from the guest out to the physical >> network.) >> >> >> but the operation "define" on virsh accepted without error the xml i've >> done, but the dumpxml doesn't show them, so, it doesn't work at all. >> I've followed the instruction that said the manual "Application >> development guide - a guide to application development with libvirt" on page >> 71. >> >> >> >> That document is targeted towards software developers writing applications >> to manage virtual guests using the libvirt API, not for system >> administrators who are configuring virtual guests manually (or, more likely, >> by using software that uses the libvirt API, eg virsh or virt-manager). It >> likely contains more detail and less overview than what you need, which is >> probably what led to the confusion. I've included a pointer to a wiki page >> below that is more the appropriate level of information (Justin, if the docs >> you were working on are somewhere easily accessible, you can point to those >> instead.) >> >> >> >> The whole definition of the VM is attached, but the related to interface >> is: >> <interface type='bridge'> >> <source bridge='br0'/> >> <target dev='vnet0'/> >> <protocol> >> <ip address="192.168.1.41" prefix="24"/> >> <route gateway="192.168.1.10"/> >> </protocol> >> <bridge stp="off" delay="0.01"> >> <interface type="ethernet" name="eth12"> >> <protocol> >> <ip address="192.168.1.41" prefix="24"/> >> <route gateway="192.168.1.10"/> >> </protocol> >> </interface> >> </bridge> >> </interface> >> >> >> >> You are confusing the guest interface config (that is part of the guest's >> XML config, and described above), and host physical interface config (that >> is not in any way associated with guest config, but instead used to >> configure the host's physical network interfaces, bridges, bonds, and vlan >> interfaces; and nearly as important, is not supported on Ubuntu anyway). >> >> Actual guest interface config is *much* simpler than that. >> >> >> >> >> But after define, the dumpxml only show this part: >> <interface type='bridge'> >> <mac address='52:54:00:3c:92:9a'/> >> <source bridge='br0'/> >> </interface> >> >> >> >> That's because everything else you've put in your guest config is not >> actually a part of a valid guest interface config, and so is ignored. >> >> >> >> >> I would appreciate any help to solve the question. The problem I've is I >> lost the access to the guest every time the libvirt change the name of the >> interface eth1,2,3, etc. >> >> >> In this case I think it may be useful to start over. You've somehow been >> tricked into misunderstanding the way that network configuration in libvirt >> works, confusing host interface config with guest interface config, and the >> resulting config bears little resemblance to what would actually work. >> >> == >> The first step is to read this page: >> >> http://wiki.libvirt.org/page/Networking >> >> to get a basic overview. >> >> == >> Second step: you should decide which of the three methods of connecting to >> guests to the physical network you want to use: >> >> 1) NATed >> 2) routed >> 3) bridged >> >> The simplest is (1), and that will probably work just fine as long as 1) >> you don't need incoming connections to the guests, and 2) you're not >> concerned about getting the last fraction of an ounce of performance out of >> the network. >> >> Depending on which of those 3 you decide to use: >> >> == >> Third step: follow the instructions to either create a bridge interface on >> the host connected to a physical network interface (for (3)), or to create a >> libvirt virtual network using one of the two modes ((1) or (2) - actually >> the stock libvirt config already has a "default" network that is probably >> adequate for (1), and unless you understand IP routing very well, you'll >> likely want to avoid (2)). >> >> Once the physical network is setup: >> >> == >> Fourth step: define your guests. In each guest, you'll have an <interface> >> element, but it will be much simpler than what you've previously tried. For >> bridged mode, the interface section would look like this: >> >> >> <interface type='bridge'> >> <source bridge='br0'/> >> </interface> >> >> Depending on the guest, you may want to specify what type of hardware is >> presented to the guest (default is rtl, which almost always works, but is >> not the best perfrorming), in particular, if the guest supports virtio-net, >> you'll want to add this into the <interface> element: >> >> <model type='virtio'/> >> >> For NATed or routed mode, the guest's interface definition would look like >> this: >> >> <interface type='network'> >> <source network='default'/> >> </interface> >> >> (replace "default" with the name of the network you've defined, if you've >> defined your own). In this case also, you can add a <model type='...'/> if >> appropriate. >> >> Note that I don't put anything in the <interface> element about MAC >> address, target device, or bus address. Those are all setup automatically by >> libvirt the first time you start the guest, and it will make sure that the >> guests don't interfere with each other. >> >> == >> Fifth step: only if you need a particular guest to have a particular fixed >> IP address, either 1) edit the network config file on the guest, 2) add a >> static host to the libvirt virtual network definition (giving the MAC >> address provided to the guest by libvirt during the initial guest startup) >> with a <host> element added to the <dhcp> element of the network; details >> are here: >> >> http://www.libvirt.org/formatnetwork.html >> >> or 3) if you're using the bridged method to connect to the physical >> network, again either configure the guest's own config file, or add a static >> host definition to the dhcp server running on your physical network. >> >> If you still have problems don't hesitate to send another mail (although, >> as I said at the top, sending to libvirt-users(a)redhat.com may yield >> better results than libvir-list(a)redhat.com) >> > >

3 5

[libvirt-users] How to retrieve Memory stat of the KVM using python binding with Libvirt API
by arpita k 10 Feb '11

10 Feb '11

Hi , I am using ubuntu host machine and KVM is present as guest in that machine.Two virtual machines are there.I am using Python binding to query on the hypervisor and extract the VM related information.I like to print memory stat of the VM like SWAP_IN , SWAP_OUT,free memory, mem cached etc.So, the code, I have written for this, is : #!/usr/local/bin/python import libvirt import sys conn = libvirt.openReadOnly(None) if conn == None: print "Failed to open connection to the hypervisor" sys.exit(1) try: domIds = conn.listDomainsID() for id in domIds: dom0 = conn.lookupByID(id) try: print "memory stats: %s"% dom0.memoryStats().values() except: print 'Failed to find the memory data of this domain' sys.exit(1) except: print 'Failed to find the main domain' sys.exit(1) But this code 'dom0.memoryStats().values()' is returning empty list.I need to collect more information regarding the memory .How should I proceed? Regards, Arpita

1 0