- Devel - libvirt List Archives

[libvirt] [PATCH 0/7] Support cgroups in QEMU driver
by Daniel P. Berrange 20 Jul '09

20 Jul '09

This series of patches introduces cgroups support to the QEMU driver. At this time it uses the 'devices' controller to whitelist block devices for QEMU guests to prevent unauthorized access. It uses the 'cpu_shares' controller to allow schedular tunables on a per guest basis. In the future we should use the 'memory' controller to enforce the limit set in the balloon driver ie if the guest does not honour the balloon request, then force the guest into swap, avoiding host overcommit. There is also some refactoring of the cgroups code to remove the assumption that libvirtd is starting in the root cgroup, remove the requirement that all controllers be active, and allow for use in non-privileged drivers. Daniel P. Berrange (7): Use enums for cgroup controller types / labels Use virFileReadAll/virFileWriteStr for key cgroup read/write helpers Make cgroups a little more efficient Place every QEMU guest in a private cgroup Implement schedular tunables API using cgroups Use cgroups for block device whitelisting in QEMU guests Refactor cgroups to allow a group per driver to be managed directly src/cgroup.c | 861 ++++++++++++++++++++++++++-------------------- src/cgroup.h | 28 +- src/libvirt_private.syms | 1 + src/lxc_conf.h | 2 + src/lxc_controller.c | 19 +- src/lxc_driver.c | 25 +- src/qemu_conf.h | 2 + src/qemu_driver.c | 355 ++++++++++++++++++- src/util.c | 27 ++- src/util.h | 2 + 10 files changed, 904 insertions(+), 420 deletions(-)

2 17

[libvirt] changing files in /var/lib/libvirt/images
by Gene Czarcinski 20 Jul '09

20 Jul '09

Assume I am working with virtual disk image files in /var/lib/libvirt/images such as removing a file, defining a new file (with qemi-img create or qemu-img convert), or simply copying a file ("cp") from elsewhere such as VMware. When I do this, Guest definition or adding hardware to an existing guest under virt- manager does not "see" the file change. That is, until I terminate virt- manager and restart /etc/init.d/libvirtd. Is there some better/easier/less-disruptive means of notifying "whatever" that the list of files has changed? I noticed that /etc/init.d/libvirtd force-reload has no effect. I also notice that restarting libvirtd and/or terminating and restarting virt- manager while there are guests running really screws things up. If such a "means" does not exist, should one be created? [should this be an RFE in bugzilla?] Gene

2 2

[libvirt] Create a new VM from an existent image?
by Jun Koi 19 Jul '09

19 Jul '09

Hi, To use libvirt, I am strying to use virt-install to create a new VM from an existent (KVM) VM image. I did the following steps: 1) run libvirtd with "libvirtd -d" (I use the latest libvirt git tree) 2) Run virt-install (0.400.3) with my existent image in img/img.winxp # virt-install --import -n winxp -f img/img.xp3 --vnc --hvm --accelerate --ram 800 However, this step returns error like: "ERROR Could not find usable default libvirt connection." Actually I am not sure that is the right way to use the new --import option with virt-install. So that might be the problem? All the code are updated, and compiled from source on Ubuntu 8.04. Any help? Thanks a lot, J

2 4

[libvirt] [PATCH] qemu: Try multiple times to open unix monitor socket
by Cole Robinson 17 Jul '09

17 Jul '09

Unlike the pty monitor (which we know exists since we scrape its path from stdout), we have no way of knowing that the unix monitor socket should exist/ be initialized. As a result, some of my KVM guests randomly fail to start on F10 host. Try to open the unix socket in a 3 second timeout loop. Ignore EACCES (path does not exist if a first time run) and ECONNREFUSED (leftover socket from a previous run hasn't been removed yet). Fixes things for me. --- src/qemu_driver.c | 22 +++++++++++++++++++++- 1 files changed, 21 insertions(+), 1 deletions(-) diff --git a/src/qemu_driver.c b/src/qemu_driver.c index e2b7acb..200718b 100644 --- a/src/qemu_driver.c +++ b/src/qemu_driver.c @@ -874,6 +874,8 @@ qemudOpenMonitorUnix(virConnectPtr conn, { struct sockaddr_un addr; int monfd; + int timeout = 3; /* In seconds */ + int ret, i; if ((monfd = socket(AF_UNIX, SOCK_STREAM, 0)) < 0) { virReportSystemError(conn, errno, @@ -885,10 +887,28 @@ qemudOpenMonitorUnix(virConnectPtr conn, addr.sun_family = AF_UNIX; strncpy(addr.sun_path, monitor, sizeof(addr.sun_path)); - if (connect(monfd, (struct sockaddr *) &addr, sizeof(addr)) < 0) { + do { + ret = connect(monfd, (struct sockaddr *) &addr, sizeof(addr)); + + if (ret == 0) + break; + + if (errno == EACCES || errno == ECONNREFUSED) { + /* EACCES : Socket may not have shown up yet + * ECONNREFUSED : Leftover socket hasn't been removed yet */ + continue; + } + virReportSystemError(conn, errno, "%s", _("failed to connect to monitor socket")); goto error; + + } while ((++i <= timeout*5) && (usleep(.2 * 1000000) <= 0)); + + if (ret != 0) { + virReportSystemError(conn, errno, "%s", + _("monitor socket did not show up.")); + goto error; } if (qemudOpenMonitorCommon(conn, driver, vm, monfd, reconnect) < 0) -- 1.6.0.6

4 5

[libvirt] adding scsi disk failed (code=9)
by Massimo Canonico 17 Jul '09

17 Jul '09

Hi, I'm newbie and sorry if this question was already post in this ML (is there any search engine for the ML archive?). I'm trying to attach a volume in my instance. The instance is running and the volume is available, but i got this error: ---- libvirt: operation failed: adding scsi disk failed (code=9) virDomainAttachDevice() failed (err=-1) XML=<disk type='block'><driver name='phy'/><source dev='/dev/etherd/e0.8'/><target dev='sdb'/></disk> doAttachVolume() failed error=1 ---- I'm not sure where I can start to debug this error. I'm using Fedora 11, x86_64, with these packages [root@minicloud eucalyptus]# rpm -qa | grep libv libvirt-devel-0.6.2-12.fc11.x86_64 libvorbis-1.2.0-7.fc11.x86_64 libvirt-0.6.2-12.fc11.x86_64 libvirt-python-0.6.2-12.fc11.x86_64 libvolume_id-141-3.fc11.x86_64 May you help me, please? Thanks in advance, M

4 5

[libvirt] [libvirt 0.6.4] Unable to set /domain/devices/serial[@port=1] as console?
by Charles Duffy 17 Jul '09

17 Jul '09

When I do a "virsh define" with the following: <serial type='file'> <source path='/path/to/serial.log'/> <target port='0'/> </serial> <serial type='pty'> <target port='1'/> </serial> <console type='pty'> <target port='1'/> </console> The result is this: <serial type='file'> <source path='/path/to/serial.log'/> <target port='0'/> </serial> <serial type='pty'> <target port='1'/> </serial> <console type='file'> <source path='/path/to/serial.log'/> <target port='0'/> </console> ...a markedly different configuration. Is this the result of a bug, or a known behavior I've failed to catch in the docs? Thanks!

2 1

[libvirt] Tentative schedule for 0.7.0
by Daniel Veillard 17 Jul '09

17 Jul '09

Basically I would like to freeze development at the end of next week and push at the end of the month. Since I plan to be away on 31, I will probably make the release on the 30th, but I think we can still apply new features until Friday next week. To try to help w.r.t. Fedora 12 schedule I will make a 0.7.0 beta release probably on the week-end of the 25 and push it in rawhide for parallel testing, Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel(a)veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/

1 0

[libvirt] One line logging fix
by Dave Allan 17 Jul '09

17 Jul '09

I noticed that I was incorrectly logging a debug message as an error in a success path in the node device code. Here's the fix. >From 0c65b2d1d90bf0d469ffe6d8a20a5566b359a61e Mon Sep 17 00:00:00 2001 From: David Allan <dallan(a)redhat.com> Date: Fri, 17 Jul 2009 09:57:40 -0400 Subject: [PATCH 1/1] Remove spurious log message A success path in part of the node device code contained a log message with severity ERROR; this fix makes it DEBUG. --- src/node_device_hal_linux.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/node_device_hal_linux.c b/src/node_device_hal_linux.c index ddb1eef..a119404 100644 --- a/src/node_device_hal_linux.c +++ b/src/node_device_hal_linux.c @@ -50,7 +50,7 @@ static int open_wwn_file(const char *prefix, /* fd will be closed by caller */ if ((*fd = open(wwn_path, O_RDONLY)) != -1) { - VIR_ERROR(_("Opened WWN path '%s' for reading"), + VIR_DEBUG(_("Opened WWN path '%s' for reading"), wwn_path); } else { VIR_ERROR(_("Failed to open WWN path '%s' for reading"), -- 1.6.0.6

2 1

[libvirt] Guest shutdown script
by Ján ONDREJ (SAL) 17 Jul '09

17 Jul '09

Hello, when host server is shutting down, it's goot to shutdown also guests, not only kill them. Host shutdown can be touched by an admin, but also can be started by an UPS daemon or similar action. Then kill of guests is not a best solution. I have my own script, which can shutdown guest automatically: http://www.salstar.sk/pub/salpack/etc/rc.d/init.d/virt-shutdown You need to start this rc script after installation, otherwise it will not do anything. Another init script with save/restore ability is here: http://marc.info/?l=kvm&m=124775655529836&w=4 I'm not using this so I don't know, if this will work well on Fedora. A good solution can also be, if libvirt init script can shutdown guests autoamtically. It starts them, when system is booting, but does not stop them. I know, that libvirt sometimes need to be restarted without touching guests, but you also can avoid this, if you shutdown guest only if one of: - current runlevel is 0 (halt) or 6 (reboot). You can get current runlevel by running: runlevel | cut -d" " -f2 - if "service libvirt shutdown" is called, which is not very systematic, but can work well Just please do not create a new package only for guest shutdowns. :-) I am not a member of this list, so please cc me. SAL

1 0

[libvirt] How to pass not xml parameters from Libvirt to Xen in xml definition
by Alexandre CABROL PERALES 17 Jul '09

17 Jul '09

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi everybody, i need to define cpuid parameter for a specific Xen Virtual Machine. This parameters doesn't exist in libvirt definition. So how can i add this "unknow" parameter in virtual machine's xml definition to be able to manage it from virt-manager? regards, - -- Alexandre CABROL PERALES - -- Ingenieur Securite des Systemes d'Information Mob. : 06.98.82.03.06 Mail : alexandre.cabrol(a)artal.fr Key fingerprint = 1E6B B8DF 5001 A6A8 E057 9D31 7B3B EAB1 4AE4 8953 - -- ARTAL Technologies Rue Pierre-Gilles de Gennes Ens."La Rue", Bat. 9, BP 38138 31681 Labege cedex Tel. : 05.61.00.39.30 Fax : 05.61.00.20.43 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkpgSWsACgkQezvqsUrkiVMqiwCgkigm/WZ5TPToQ7RzdY0R6UJz tEAAoKIcdZ2fO5PdqZdcrzuTj6Ev4yh4 =8rlJ -----END PGP SIGNATURE-----

2 2