- Devel - libvirt List Archives

[Libvir] [patch 1/3] Move libvirt_qemud to /usr/sbin
by Mark McLoughlin 23 Feb '07

23 Feb '07

Move libvirt_qemud to /usr/sbin since it is now "designed to be executed by hand" ... i.e. we're adding an initscript for it rather than just having libvirt start it automatically. Signed-off-by: Mark McLoughlin <markmc(a)redhat.com> Index: libvirt/qemud/Makefile.am =================================================================== --- libvirt.orig/qemud/Makefile.am +++ libvirt/qemud/Makefile.am @@ -2,7 +2,7 @@ INCLUDES = @LIBXML_CFLAGS@ -libexec_PROGRAMS = libvirt_qemud +sbin_PROGRAMS = libvirt_qemud libvirt_qemud_SOURCES = qemud.c internal.h protocol.h \ driver.c driver.h \ Index: libvirt/src/Makefile.am =================================================================== --- libvirt.orig/src/Makefile.am +++ libvirt/src/Makefile.am @@ -1,7 +1,7 @@ ## Process this file with automake to produce Makefile.in INCLUDES = -I$(top_builddir)/include -I@top_srcdir@/include @LIBXML_CFLAGS@ -I@top_srcdir@/qemud \ - -DBINDIR=\""$(libexecdir)"\" -DLOCALEBASEDIR=\""$(datadir)/locale"\" \ + -DBINDIR=\""$(libexecdir)"\" -DSBINDIR=\""$(sbindir)"\" -DLOCALEBASEDIR=\""$(datadir)/locale"\" \ -DLOCAL_STATE_DIR=\""$(localstatedir)"\" \ -DGETTEXT_PACKAGE=\"$(PACKAGE)\" DEPS = libvirt.la Index: libvirt/src/qemu_internal.c =================================================================== --- libvirt.orig/src/qemu_internal.c +++ libvirt/src/qemu_internal.c @@ -99,8 +99,8 @@ static const char * qemuFindServerPath(void) { static const char *serverPaths[] = { - BINDIR "/libvirt_qemud", - BINDIR "/libvirt_qemud_dbg", + SBINDIR "/libvirt_qemud", + SBINDIR "/libvirt_qemud_dbg", NULL }; int i; --

1 0

[Libvir] [patch 0/3] Add qemud init script
by Mark McLoughlin 23 Feb '07

23 Feb '07

Hey, In order to run autostart networks and guests at boot, we need an init script. Here she comes ... Cheers, Mark. --

1 0

[Libvir] PATCH: Autodetect QEMU version
by Daniel P. Berrange 23 Feb '07

23 Feb '07

When I tested out the libvirt 0.2.0 in Fedora rawhide I discovered that the syntax for specifying VNC display numbers changes in QEMU 0.9.0. Previously you would use -vnc 7 Now you need to use -vnc :7 For current Fedora RPM I just hacked in a workaround to always add ':', but we need to be able to detect the version properly. So I'm attaching a patch which spawns '/usr/bin/qemu' and groks its command line help to extract the version number. It then uses appropriate VNC argument syntax based on the version. At the same time I also check for whether -no-kqemu is available so we can turn off KQEMU support if it is not available. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|

2 4

Re: [Libvir] [patch 3/9] Re-factor chunks of qemuds config handling
by Daniel P. Berrange 23 Feb '07

23 Feb '07

On Thu, Feb 22, 2007 at 11:04:37AM +0000, Mark McLoughlin wrote: > The easiest way to explain where I'm coming from with > this patch is to look at the current qemudLoadConfigXML(). > > It's used in three places: > > 1) Loading at startup - you want (a) and (b), but you > want to check whether the name matches the filename > before doing (b) > > 2) Create() - you want (a) and (b) > > 3) Define() - you want (a), (b) and (c) > > So, mostly I'm just splitting the function into three > and making the logic above much more obvious. It should > also make some of the autostart code much more > straightforward. Hard to tell from the raw patch exactly how it fits together, but the principles of the patch all sound good to me - centralizing alot of the pointer munging into one place & generally tidying stuff up. BTW, your mailer alias for this list is wrong - its generating a bogus reply to against libvirt hacker dudes <libvirt-list(a)redhat.com> Instead of libvirt hacker dudes <libvir-list(a)redhat.com> Note, the extra 't' in the former - which makes replying to your mails a little error prone - several went to dev null... Dan -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|

2 1

[Libvir] Autostarting guests and networks
by Mark McLoughlin 23 Feb '07

23 Feb '07

Hey, We need some way to have libvirtd automatically start guests and networks. We had said we should have autostart directories containing config files of guests/networks which should be autostarted. One problem I see with that is that we'd need a new API to define autostart configs. So, I suggest we add an "autostart" flag to the toplevel element of the configs. We could discuss this one at length and I, for one, don't like using the XML format as an API like this, but ... comments? Cheers, Mark.

6 7

Re: [Libvir] [patch 0/9] Revised autostart patches
by Daniel P. Berrange 23 Feb '07

23 Feb '07

On Thu, Feb 22, 2007 at 11:04:08AM +0000, Mark McLoughlin wrote: > Hey, > Here's the autostart patches again. If you're just > looking for what's changed, it's probably fine to just look > at patch 3 (some heavy re-factoring to make the autostart > stuff easier) and patch 7 (maintains symlinks in the autostart > dirs instead of config files). Aside from the note about 'virsh list' command changes, it all looks good to me. Unless anyone else has objections I'd say commit this to CVS - I want to re-do the handling of the QEMU monitor to detect errors at startup and would rather work with your big refactoring already applied.. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|

1 0

Re: [Libvir] [patch 1/9] Merge the active and inactive network/guest lists
by Daniel P. Berrange 23 Feb '07

23 Feb '07

On Thu, Feb 22, 2007 at 11:04:20AM +0000, Mark McLoughlin wrote: > Merge the ->activevms and ->inactivevms into a single > ->vms list where each VM has a "active" flag in order > to make things easier to manage. That looks good now with the qemudIsActiveVM() helper function. Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|

1 0

[Libvir] PATCH: Don't leave a 'braindead' Xen VM when device hotplug fails
by Daniel P. Berrange 23 Feb '07

23 Feb '07

Because of the way Xen guest creation works, the virDomainCreateLinux implementation for Xen is really a three stage process - first we create the raw domain, then we have to wait for device hotplug to complete, finally unpausing the domain. Currently if the device hotplug fails we error out, leaving a 'braindead' VM lying around in the paused state. This is rather confusing for users leading to people thinking everything was OK, and thus unpausing the guest manually, then filing bug reports when they find they've no devices in the guest VM ! The xm tool by comparison will tear down the VM if the device hotplug fails so the user doesn't see the paused domain after the failure. The attached patch re-arranges the Xen driver for creating new domains to do the same kind of thing - upon any failure we explicitly destroy the halfdead domain Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|

2 2

[Libvir] [PATCH]Guard for Dom0 by virsh domain stop commands
by Kazuki　Mizushima 23 Feb '07

23 Feb '07

Hi, I think Domain0 should NOT stop by virsh domain stop commands. (like shutdown, reboot, destroy). But virsh shudown and reboot commands are still available for Dom0. This patch intends to disable these two commands for Dom0. Internal processing of these two commands is following 1)Send shutdown or reboot to Xend (Guard of xend) 2)If 1) is failed, shutdown/reboot is executed by xenstore. (Not Guard) This patch removes 2) ,that is as same as destroy processing. Signed-off-by: Kazuki Mizushima <mizushima.kazuk(a)jp.fujitsu.com> Thanks, Kazuki Mizushima Index: xs_internal.c (libvirt-0.2.0) ---------------------------------------------------------------------- --- xs_internal.c 2007-02-15 01:11:54.000000000 +0900 +++ xs_internal.c.guard 2007-02-22 00:31:55.000000000 +0900 @@ -54,8 +54,8 @@ static virDriver xenStoreDriver = { xenStoreDomainLookupByName, /* domainLookupByName */ NULL, /* domainSuspend */ NULL, /* domainResume */ - xenStoreDomainShutdown, /* domainShutdown */ - xenStoreDomainReboot, /* domainReboot */ + NULL, /* domainShutdown */ + NULL, /* domainReboot */ NULL, /* domainDestroy */ xenStoreDomainGetOSType, /* domainGetOSType */ xenStoreDomainGetMaxMemory, /* domainGetMaxMemory */ ----------------------------------------------------------------------

3 4

[Libvir] [patch 9/9] Actually autostart guests and networks
by Mark McLoughlin 22 Feb '07

22 Feb '07

Add qemudAutostartConfigs() to autostart guests and networks. Signed-off-by: Mark McLoughlin <markmc(a)redhat.com> Index: libvirt/qemud/conf.c =================================================================== --- libvirt.orig/qemud/conf.c +++ libvirt/qemud/conf.c @@ -1972,11 +1972,49 @@ int qemudScanConfigDir(struct qemud_serv return 0; } +static +void qemudAutostartConfigs(struct qemud_server *server) { + struct qemud_network *network; + struct qemud_vm *vm; + + network = server->networks; + while (network != NULL) { + struct qemud_network *next = network->next; + + if (network->autostart && + !qemudIsActiveNetwork(network) && + qemudStartNetworkDaemon(server, network) < 0) + qemudLog(QEMUD_ERR, "Failed to autostart network '%s'", + network->def->name); + + network = next; + } + + vm = server->vms; + while (vm != NULL) { + struct qemud_vm *next = vm->next; + + if (vm->autostart && + !qemudIsActiveVM(vm) && + qemudStartVMDaemon(server, vm) < 0) + qemudLog(QEMUD_ERR, "Failed to autostart VM '%s'", + vm->def->name); + + vm = next; + } +} + /* Scan for all guest and network config files */ int qemudScanConfigs(struct qemud_server *server) { if (qemudScanConfigDir(server, server->configDir, server->autostartDir, 1) < 0) return -1; - return qemudScanConfigDir(server, server->networkConfigDir, server->networkAutostartDir, 0); + + if (qemudScanConfigDir(server, server->networkConfigDir, server->networkAutostartDir, 0) < 0) + return -1; + + qemudAutostartConfigs(server); + + return 0; } /* Simple grow-on-demand string buffer */ --

1 0