December 2008 - Devel - libvirt List Archives

[libvirt] broken pipe?
by Itamar Heim 08 Dec '08

08 Dec '08

Hi, I have a problem with a small python script (using fedora 10 with its shipped rpm's - libvirt-0.4.6-3.fc10.x86_64, libvirt-python-0.4.6-3.fc10.x86_64) I'm running a simple stupid libvirt python script. I happen to want to use SDL, and it fails on "Could not initialize SDL - exiting". While I need to solve this error (i.e., find out in which version this patch was included to allow me to specify the DISPLAY http://www.mail-archive.com/libvir-list@redhat.com/msg08764.html) I am more troubled by the fact that running the script again gives a socket/pipe error until I restart libvirtd: Run #1: ./test_libvirt.py libvir: QEMU error : internal error QEMU quit during console startup Could not initialize SDL - exiting Traceback (most recent call last): File "./test_libvirt.py", line 37, in <module> vm = c.createLinux(x, 0) File "/usr/lib64/python2.5/site-packages/libvirt.py", line 892, in createLinux if ret is None:raise libvirtError('virDomainCreateLinux() failed', conn=self) libvirt.libvirtError: internal error QEMU quit during console startup Run #2: ./test_libvirt.py libvir: Remote error : socket closed unexpectedly Traceback (most recent call last): File "./test_libvirt.py", line 37, in <module> vm = c.createLinux(x, 0) # what a lousy function name File "/usr/lib64/python2.5/site-packages/libvirt.py", line 892, in createLinux if ret is None:raise libvirtError('virDomainCreateLinux() failed', conn=self) libvirt.libvirtError: socket closed unexpectedly libvir: Remote error : Broken pipe Code: #!/usr/bin/env python import libvirt c = libvirt.open('qemu:///system') x = """<domain type='kvm'> <name>yossi2</name> <memory>512000</memory> <currentMemory>512000</currentMemory> <vcpu>1</vcpu> <os> <type arch='i686' machine='pc'>hvm</type> <boot dev='hd'/> </os> <features> <acpi/> </features> <clock offset='utc'/> <devices> <emulator>/usr/bin/qemu-kvm</emulator> <disk type='file' device='disk'> <source file='/mnt/sda2/images/winxp-acpi.vmdk'/> <target dev='hda' bus='ide'/> </disk> <input type='tablet' bus='usb'/> <input type='mouse' bus='ps2'/> <graphics type='sdl'/> </devices> </domain>""" vm = c.createLinux(x, 0) Thanks, Itamar

4 10

[libvirt] FYI: Fixed multiple test driver crashes/flaws
by Daniel P. Berrange 07 Dec '08

07 Dec '08

The virt-install test suite exposed a number of bugs in the test driver storage APIs, partly due to my threading changes. There was one deadlock (calling unneccessary PoolRefresh op), one crash (deferencing a variable we just set to NULL), and integer long long overflow on 32-bit, and missing return value I've committed the following patch which fixes these and makes virt=install test suite work again Daniel diff -u -p -r1.103 -r1.106 --- src/test.c 4 Dec 2008 21:00:50 -0000 1.103 +++ src/test.c 7 Dec 2008 20:45:28 -0000 1.106 @@ -191,7 +191,7 @@ static const char *defaultPoolXML = " </target>" "</pool>"; -static const unsigned long long defaultPoolCap = (100 * 1024 * 1024 * 1024ul); +static const unsigned long long defaultPoolCap = (100 * 1024 * 1024 * 1024ull); static const unsigned long long defaultPoolAlloc = 0; static int testStoragePoolObjSetDefaults(virStoragePoolObjPtr pool); @@ -2323,9 +2323,6 @@ no_memory: return -1; } -static int -testStoragePoolRefresh(virStoragePoolPtr obj, - unsigned int flags ATTRIBUTE_UNUSED); static int testStoragePoolStart(virStoragePoolPtr pool, @@ -2350,8 +2347,6 @@ testStoragePoolStart(virStoragePoolPtr p goto cleanup; } - if (testStoragePoolRefresh(pool, 0) == 0) - goto cleanup; privpool->active = 1; ret = 0; @@ -2506,6 +2501,7 @@ testStoragePoolBuild(virStoragePoolPtr p _("storage pool '%s' is already active"), pool->name); goto cleanup; } + ret = 0; cleanup: if (privpool) @@ -3001,10 +2997,10 @@ testStorageVolumeCreateXML(virStoragePoo privpool->def->allocation); privpool->volumes.objs[privpool->volumes.count++] = privvol; - privvol = NULL; ret = virGetStorageVol(pool->conn, privpool->def->name, privvol->name, privvol->key); + privvol = NULL; cleanup: virStorageVolDefFree(privvol); -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

1 0

[libvirt] PATCH: Workaround limits.h bug on Debian
by Daniel P. Berrange 05 Dec '08

05 Dec '08

The Debian Etch /usr/include/limits.h has an #endif in the wrong place, making it fail to define ULLONG_MAX. This works around the brokenness Daniel diff -u -r1.32 storage_conf.c --- storage_conf.c 4 Dec 2008 22:00:14 -0000 1.32 +++ storage_conf.c 5 Dec 2008 14:17:23 -0000 @@ -43,6 +43,11 @@ #include "util.h" #include "memory.h" +/* Work around broken limits.h on debian etch */ +#if defined __GNUC__ && defined _GCC_LIMITS_H_ && ! defined ULLONG_MAX +# define ULLONG_MAX ULONG_LONG_MAX +#endif + #define virStorageLog(msg...) fprintf(stderr, msg) VIR_ENUM_IMPL(virStoragePool, -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

2 1

[libvirt] [PATCH] Makefile.am compile fix
by Paolo Borelli 05 Dec '08

05 Dec '08

Hi, also filed as https://bugzilla.redhat.com/show_bug.cgi?id=474211 but I have been asked to send it to the list. The attached patches fixes Makefile.am and let's me compile libvirt. Ciao Paolo

1 0

[libvirt] how to install updated python bindings
by ente linux 05 Dec '08

05 Dec '08

hi i was trying to install libvirt 0.5 on my centos 5 machine which by default have libvirt 0.4. But after installing from the source of libvirt, still the output of i get from python remains that of 0.4 >>>import libvirt >>> libvirt.getVersion() 4006 how could i install the new version... Do i need to give any option while configuring or making. thanks and regards visco

5 4

[libvirt] [RFC PATCH] Let network bridge MTU to be configured on XML
by Eduardo Habkost 04 Dec '08

04 Dec '08

Hi, Below is an incomplete patch for letting the MTU of a libvirt virtual network to be configured on XML. This implementation has an issue: it works when trying to set the MTU below 1500, but the kernel bridge code doesn't let us to set the bridge MTU above 1500 if there is no interface attached to the bridge yet. To allow setting the bridge MTU higher than 1500, we need to save the configured value somewhere and set it only when attaching tap interface to it. To solve that, I propose changing the way the network driver interface expose things for the domain network code. Today we do the following (src/qemu_conf.c): virNetworkPtr network; virDomainNetDefPtr net; char *brname; int tapfd; brname = virNetworkGetBridgeName(network); brAddTap(brctl, brname, &net->ifname, &tapfd))) With this code, the only information that can flow between the network driver and the bridge code is the bridge name. I was going to suggest having something like this: virNetworkAddTap(network, &net->ifname, &tapfd); However, we wouldn't be able to return an FD when using the 'remote' network driver. But we can split the creation of the tap interface and attaching it to the network bridge, like this: brNewTap(brctl, &net->ifname, &tapfd); virNetworkAttachInterface(network, net->ifname); This way, the network driver code can do anything needed to attach the new tap interface to the virtual network, including but not limited to MTU setting. Another alternative could be creating a virNetworkGetBridgeMtu() function. That would solve our problem for the MTU settings, but in my opinion, attaching interfaces to the network is a task for the network driver. Having the qemu code calling brAddTap() directly was just a shortcut we won't be able to use anymore. What do you think? --- src/bridge.c | 19 +++++++++++++++++++ src/bridge.h | 4 ++++ src/libvirt_sym.version.in | 1 + src/network_conf.c | 5 +++++ src/network_conf.h | 2 ++ src/network_driver.c | 10 ++++++++++ 6 files changed, 41 insertions(+), 0 deletions(-) diff --git a/src/bridge.c b/src/bridge.c index 13d81bc..ed9023b 100644 --- a/src/bridge.c +++ b/src/bridge.c @@ -452,6 +452,25 @@ brAddTap(brControl *ctl, return errno; } + +/** + * brSetMtu + * @ctl: bridge control pointer + * @bridge: the bridge name + * @mtu: the MTU value to be set + * + * Sets the MTU of a bridge. + * + * Returns 0 in case of success or an errno code in case of failure. + */ +int +brSetMtu(brControl *ctl, + const char *bridge, + int mtu) +{ + return ifSetMtu(ctl, bridge, mtu); +} + /** * brSetInterfaceUp: * @ctl: bridge control pointer diff --git a/src/bridge.h b/src/bridge.h index 7eb6166..194e96f 100644 --- a/src/bridge.h +++ b/src/bridge.h @@ -58,6 +58,10 @@ int brDeleteInterface (brControl *ctl, const char *bridge, const char *iface); +int brSetMtu (brControl *ctl, + const char *bridge, + int mtu); + int brAddTap (brControl *ctl, const char *bridge, char **ifname, diff --git a/src/libvirt_sym.version.in b/src/libvirt_sym.version.in index ec9ff3d..668b902 100644 --- a/src/libvirt_sym.version.in +++ b/src/libvirt_sym.version.in @@ -274,6 +274,7 @@ LIBVIRT_PRIVATE_@VERSION@ { global: /* bridge.h */ brAddBridge; + brSetMtu; brAddInterface; brAddTap; brDeleteBridge; diff --git a/src/network_conf.c b/src/network_conf.c index 5add62e..b7ec87e 100644 --- a/src/network_conf.c +++ b/src/network_conf.c @@ -295,6 +295,7 @@ virNetworkDefParseXML(virConnectPtr conn, { virNetworkDefPtr def; char *tmp; + long mtu; if (VIR_ALLOC(def) < 0) { virNetworkReportError(conn, VIR_ERR_NO_MEMORY, NULL); @@ -332,6 +333,8 @@ virNetworkDefParseXML(virConnectPtr conn, /* Parse bridge information */ def->bridge = virXPathString(conn, "string(./bridge[1]/@name)", ctxt); + if (virXPathLong(conn, "number(./bridge[1]/@mtu)", ctxt, &mtu) == 0) + def->bridgeMtu = mtu; tmp = virXPathString(conn, "string(./bridge[1]/@stp)", ctxt); def->stp = (tmp && STREQ(tmp, "off")) ? 0 : 1; VIR_FREE(tmp); @@ -570,6 +573,8 @@ char *virNetworkDefFormat(virConnectPtr conn, virBufferAddLit(&buf, " <bridge"); if (def->bridge) virBufferEscapeString(&buf, " name='%s'", def->bridge); + if (def->bridgeMtu > 0) + virBufferVSprintf(&buf, " mtu='%d'", def->bridgeMtu); virBufferVSprintf(&buf, " stp='%s' forwardDelay='%ld' />\n", def->stp ? "on" : "off", def->delay); diff --git a/src/network_conf.h b/src/network_conf.h index 2614e47..a30cc8d 100644 --- a/src/network_conf.h +++ b/src/network_conf.h @@ -61,6 +61,8 @@ struct _virNetworkDef { char *name; char *bridge; /* Name of bridge device */ + int bridgeMtu; /* MTU of the bridge */ + char *domain; unsigned long delay; /* Bridge forward delay (ms) */ int stp : 1; /* Spanning tree protocol */ diff --git a/src/network_driver.c b/src/network_driver.c index 96d3ddf..10bdd10 100644 --- a/src/network_driver.c +++ b/src/network_driver.c @@ -684,6 +684,16 @@ static int networkStartNetworkDaemon(virConnectPtr conn, return -1; } + if (network->def->bridgeMtu> 0) { + if ((err = brSetMtu(driver->brctl, network->def->bridge, network->def->bridgeMtu))) { + networkReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, + _("cannot set MTU of bridge '%s' to %d: %s"), + network->def->bridge, network->def->bridgeMtu, + strerror(err)); + goto err_delbr; + } + } + if (brSetForwardDelay(driver->brctl, network->def->bridge, network->def->delay) < 0) goto err_delbr; -- 1.5.5.GIT -- Eduardo

3 6

[libvirt] Release of libvirt-0.5.1
by Daniel Veillard 04 Dec '08

04 Dec '08

Mostly a bug fix one, as promised when 0.5.0 came out :-) Available as usual from: ftp://libvirt.org/libvirt/ Some bug fixes and local improvements: + Portability: - fix missing dep in spec file - fix compilation with new NUMA libraries (Daniel Berrange) - udev compatibility for RHEL (Chris Lalancette) - + Documentation: - documentation copy and paste errors and typo (Cole Robinson) + Bug fixes: - add a delay in storage backend for disks to show up (Chris Lalancette) - fix parsing for CDRom device with no source (Daniel Berrange) - use xenstore to list domains to avoid some bugs (Guido Gunther) - remove a leak in xen inotify code (Daniel Berrange) - UML driver freeing of uninitialialized variable (Ron Yorston) - fix UML inotify code (Daniel Berrange) - crash when adding storage without a format (Cole Robinson) + Improvements: - use xend preferably to hypervisor call to set Xen max memory (Jim Fehlig) - allow remote://hostname/ URI for automatic probe of hypervisors (Daniel Berrange) - fix daemon configuration regression testing (Jim Meyering ) - check /usr/bin/kvm for QEmu driver init (Guido Gunther) - proper active vs. inactive differentiation (Guido Gunther) - improve MTU setting on tap interfaces (Eduardo Habkost) - increase timeout for initial QEmu monitor poll (Cole Robinson) + Cleanups: - fix improper initialisations (Jim Meyering) thanks everybody for the contributions, whatever the way ! 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] [PATCH] Add default format info for storage volumes
by Cole Robinson 04 Dec '08

04 Dec '08

The attached patch adds default format info for storage volumes. This fixes a regression in the latest release, where defining a storage volume without a format crashed libvirtd: https://bugzilla.redhat.com/show_bug.cgi?id=473071 This is similar to how we handle things with storage pools as well. This fixes the crash for me. Thanks, Cole

3 2

[libvirt] [PATCH] Increase initial qemu monitor read timeout
by Cole Robinson 04 Dec '08

04 Dec '08

See https://bugzilla.redhat.com/show_bug.cgi?id=453491 We've been getting bug reports like the above against virt-manager for a while now: installing new VMs consistently throws an error 'Timed out while reading monitor startup output'. The user can then just retry the install and it will go off without a hitch. I only just realized that the common thread between the reports is physical install media: apparently the qemu monitor isn't too responsive while it's waiting for a cold cd drive to spin up. The attached patch jacks up the timeout from 3 seconds to 10 seconds when doing the initial monitor read. I verified that this fixes the issue. Thanks, Cole

2 2

[libvirt] [PATCH] Fix a few docs errors
by Cole Robinson 04 Dec '08

04 Dec '08

The attached patch fixes a few cut and paste docs errors I've stumbled upon, and a spelling error in a virsh command output. Thanks, Cole

2 2