- Devel - libvirt List Archives

[libvirt] RFC: Libvirt FAQ
by Cole Robinson 24 Feb '10

24 Feb '10

Hi all, If you spend enough time on #virt and various virt mailing lists, there are quite a few libvirt questions that come up regularly. So, largely in an effort to save myself and everyone from repeating ourselves, I've put together a libvirt FAQ, found here: http://wiki.libvirt.org/page/FAQ Some of the really common questions that I tried to answer (paraphrased): Q) Where does libvirt store the VM XML? A) You shouldn't really care, use virsh edit/dumpxml/define/undefine Q) How do I access the qemu monitor with libvirt? A) You don't, monitor commands should be provided by libvirt APIs, take a look at this complete list... Q) Why is libvirt messing with iptables? A) Providing NAT based network access for your VMs. If you don't want it, here's how to get rid of it. I've also tried to answer some questions that I don't feel have good straightforward answers in a highly visible place, like: - What is libvirt? - What is some of the major functionality provided by libvirt? - What does it mean when someone says 'I use libvirt'? - Difference between qemu:///system and qemu:///session Feel free to make any additions or mention any feedback. At the very least, please take a glance, so when someone inevitably asks one of these questions, you know to point them to the FAQ, rather than type out an answer. If people think this is useful, we can link to it from the main libvirt.org page, and merge the current static FAQ content linked from the sidebar into the wiki page. Thanks, Cole

2 1

[libvirt] [PATCH 00/10] Various documentation updates
by Cole Robinson 24 Feb '10

24 Feb '10

Over the months I've noticed a few XML pieces that aren't documented, so I finally sat down to hammer these out. Patch #1 is actually a bug I noticed in the process of trying to document the affected attribute. Thanks, Cole Cole Robinson (10): storage: conf: Correctly calculate exabyte unit docs: storage: Document capacity/alloc 'unit' docs: storage: <volume><key> is always generated. docs: storage: Fix backingStore <format> docs docs: storage: Document <auth> element docs: storage: Document SCSI pools docs: storage: Document multipath pools docs: domain: Document <description> element docs: network: Document STP and delay attributes docs: network: Document <domain> element docs/formatdomain.html.in | 7 ++++ docs/formatnetwork.html.in | 12 ++++++- docs/formatstorage.html.in | 34 ++++++++++++++++--- docs/storage.html.in | 76 ++++++++++++++++++++++++++++++++++++++++++++ src/conf/storage_conf.c | 4 +- 5 files changed, 124 insertions(+), 9 deletions(-)

4 18

[libvirt] PATCH [0/1] Filesystem pool formatting
by David Allan 24 Feb '10

24 Feb '10

Here's an updated patch based on the feedback from DV and danpb. I haven't added support for any additional filesystems beyond what mke2fs supports. The addition of additional filesystems is easily done and better as a separate patch. Dave

4 7

[libvirt] [PATCH] virsh.c: avoid leak on OOM error path
by Jim Meyering 23 Feb '10

23 Feb '10

No one really cares if we leak memory while dying, but who knows... freeing that buffer may let us go down more gracefully. FYI, the leak is triggered when virFileReadAll succeeds (it allocates "buffer"), yet xmlNewDoc fails: if (virFileReadAll(from, VIRSH_MAX_XML_FILE, &buffer) < 0) return FALSE; doc = xmlNewDoc(NULL); if (doc == NULL) goto no_memory; >From 03c7a44e3d5b2e7c992bebc98fc8c6a7bf63881e Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Fri, 19 Feb 2010 18:03:41 +0100 Subject: [PATCH] virsh.c: avoid leak on OOM error path * tools/virsh.c (cmdCPUBaseline): Also free "buffer" upon OOM. --- tools/virsh.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/tools/virsh.c b/tools/virsh.c index dd916f3..8756a7a 100644 --- a/tools/virsh.c +++ b/tools/virsh.c @@ -7139,6 +7139,7 @@ cleanup: return ret; no_memory: + VIR_FREE(buffer); vshError(ctl, "%s", _("Out of memory")); ret = FALSE; return ret; -- 1.7.0.233.g05e1a

3 6

[libvirt] [PATCH] Create raw storage files with O_DSYNC (again)
by Jiri Denemark 22 Feb '10

22 Feb '10

Recently we introduced O_DYSNC flag when creating raw storage files to avoid filling all disk cache with dirty pages. However, the patch got lost when virStorageBackendCreateRaw was reworked using virFileOperation. Let's use O_DYSNC again. Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com> --- src/storage/storage_backend.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-) diff --git a/src/storage/storage_backend.c b/src/storage/storage_backend.c index 8b9ed5d..3742493 100644 --- a/src/storage/storage_backend.c +++ b/src/storage/storage_backend.c @@ -356,7 +356,8 @@ virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED, goto cleanup; } - if ((createstat = virFileOperation(vol->target.path, O_RDWR | O_CREAT | O_EXCL, + if ((createstat = virFileOperation(vol->target.path, + O_RDWR | O_CREAT | O_EXCL | O_DSYNC, vol->target.perms.mode, vol->target.perms.uid, vol->target.perms.gid, createRawFileOpHook, &hdata, -- 1.7.0

3 3

[libvirt] domain restore race condition
by Laine Stump 22 Feb '10

22 Feb '10

As noted in another message, the problem I was seeing is a race condition in qemudDomainRestore(), not with my modifications to qemudDmainSave(). Here's some discussion about that problem from IRC, with a question at the bottom: > <laine> Does anyone else see a failure of domain restore (immediately > after domain save? I'm very definitely seeing it on my machine with > F12+updates testing and libvirt built from unpatched sources. > <laine> It's very reproduceable - with virsh I do "save domain > filename", then "restore filename" and it pretty much always gives me > a black screen. Then I force shutdown the guest (with virt-manager) > and do "restore filename" again. Tada! It's restored and running! > [...] > <danpb> laine: possible race condition > <danpb> laine: try putting a sleep(10) before the qemuMonitorStartCPUs > in qemuDomainRestore() Dan's suggestion *did* eliminate the failures. > [...] > <danpb> laine: this sounds like the issue with libvirt prematurely > starting execution of the CPUs before QEMU has even started restoring > (or soemthing like that) > <danpb> laine: search the archives for a mail from Charles Duffy on > this subject some time ago > Here's the BZ filed by Charles Duffy https://bugzilla.redhat.com/show_bug.cgi?id=537938 It looks like he's dealing with a race condition earlier in the restore, since his solution was to wait for the migration process to terminate somewhere inside qemudStartVMDaemon(), rather than waiting until qemudStartVMDaemon() was finished (which is what it does now). Since this wait has already been done anyway by the time of Dan's sleep(10) in my test, I don't think Charles' patch would help this situation. So is there something that libvirt can wait on here to ensure proper start? Or is there a problem in qemu? (I'm still running 0.11. I'll also try upgrading to 0.12 and see if there are changes in behavior.)

1 0

[libvirt] PATCH [0/3] Filesystem pool formatting
by David Allan 22 Feb '10

22 Feb '10

I finished these patches this afternoon, and I need to do a bit more testing before I'm comfortable having these patches committed, but I'm sending them along so I can get everybody's feedback overnight. In particular, I'm wondering if my move of the pool unref call is correct. Dave

5 11

[libvirt] Failures in existing qemu domain save/restore.
by Laine Stump 21 Feb '10

21 Feb '10

After running a bunch more tests, I am now noticing failures in domain save/restore *without* my patch (ie sources built straight from git), not 100% in either case (with or without patches), and the same saved domain sometimes fails to restore and sometimes is successful. (Most of the time (but not all), when I try to restore a domain immediately after a save, it fails, but if I then force shutdown of that failed restore and restore again from the same file, it succeeds. (Note that this is all to/from a local disk - it's not related to NFS) Is there a known problem in domain save/restore? Is anyone else seeing any failures? On 02/19/2010 02:30 AM, Laine Stump wrote: > (This patch requires that you first apply the virFileOperation patches > that I sent to the list yesterday.) > > I'm hoping someone else's keen eye can see what I'm failing to see > here. > > The patch following this message makes domain save "work" on > root-squash NFS (both the libvirt header and the QEMU state are saved > to the file), but the files that it saves don't restore properly, even > when simply saving to a local directory w/o forking - the guest screen > comes up black, or reboots immediately, or some other such thing. I've > looked at the data files and the code, and can't see what difference > is causing this - at least the header written by libvirt seems to be > the same, and the part written by qemu starts in the same place and > looks similar. > > If anyone's got an idea, please let me know! I'd like to get this > figured out and a working patch submitted before the day is done. > > -- > libvir-list mailing list > libvir-list(a)redhat.com > https://www.redhat.com/mailman/listinfo/libvir-list > >

1 0

[libvirt] [PATCH] docs: add 3 missing spaces
by Dan Kenigsberg 21 Feb '10

21 Feb '10

From: danken <danken(a)redhat.com> --- docs/architecture.html.in | 2 +- docs/auth.html.in | 2 +- docs/logging.html.in | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/docs/architecture.html.in b/docs/architecture.html.in index 8eb6458..f5cc520 100644 --- a/docs/architecture.html.in +++ b/docs/architecture.html.in @@ -83,7 +83,7 @@ drivers present in driver.h: <li>xen_internal: provide the implementation of the functions possible via direct hypervisor access</li> <li>proxy_internal: provide read-only Xen access via a proxy, the proxy code - is in the <code>proxy/</code>directory.</li> + is in the <code>proxy/</code> directory.</li> <li>xm_internal: provide support for Xen defined but not running domains.</li> <li>qemu_internal: implement the driver functions for QEmu and diff --git a/docs/auth.html.in b/docs/auth.html.in index 3ca40c3..ab6c3e9 100644 --- a/docs/auth.html.in +++ b/docs/auth.html.in @@ -72,7 +72,7 @@ available. The two libvirt daemon actions available are named <code>org.libvirt. for the RO socket, and <code>org.libvirt.unix.manage</code> for the RW socket. -As an example, to allow a user <code>fred</code>full access to the RW socket, +As an example, to allow a user <code>fred</code> full access to the RW socket, while requiring <code>joe</code> to authenticate with the admin password, would require adding the following snippet to <code>PolicyKit.conf</code>. diff --git a/docs/logging.html.in b/docs/logging.html.in index 63a0b05..f4a5878 100644 --- a/docs/logging.html.in +++ b/docs/logging.html.in @@ -96,7 +96,7 @@ <li><code>x:stderr</code> output goes to stderr</li> <li><code>x:syslog:name</code> use syslog for the output and use the given <code>name</code> as the ident</li> - <li><code>x:file:file_path</code>output to a file, with the given + <li><code>x:file:file_path</code> output to a file, with the given filepath</li> </ul> In all cases the x prefix is the minimal level, acting as a filter: -- 1.6.6

2 1

[libvirt] [RFC][PATCH 2/2] Automatically recreate macvtap interface and reconnect to qemu
by Ed Swierk 19 Feb '10

19 Feb '10

This patch implements a thread for each VM that iterates through the list of macvtap interfaces, attempting to recreate each one and reconnect it to qemu. See the PATCH 0 email for discussion. --- Index: libvirt-0.7.6/src/qemu/qemu_driver.c =================================================================== --- libvirt-0.7.6.orig/src/qemu/qemu_driver.c +++ libvirt-0.7.6/src/qemu/qemu_driver.c @@ -76,6 +76,7 @@ #include "xml.h" #include "cpu/cpu.h" #include "macvtap.h" +#include "threads.h" #define VIR_FROM_THIS VIR_FROM_QEMU @@ -96,6 +97,8 @@ struct _qemuDomainObjPrivate { int *vcpupids; qemuDomainPCIAddressSetPtr pciaddrs; + + pthread_t reconnectNetBackendsThread; }; static int qemudShutdown(void); @@ -143,6 +146,8 @@ static int qemudDomainDisconnectNetBacke virDomainObjPtr vm, virDomainNetDefPtr net); +static void *qemudDomainReconnectNetBackends(void *opaque); + static struct qemud_driver *qemu_driver = NULL; @@ -2794,6 +2799,10 @@ static int qemudStartVMDaemon(virConnect if (virDomainSaveStatus(conn, driver->caps, driver->stateDir, vm) < 0) goto abort; + if (pthread_create(&priv->reconnectNetBackendsThread, NULL, + qemudDomainReconnectNetBackends, vm->def->uuid)) + goto abort; + return 0; cleanup: @@ -5627,7 +5636,7 @@ static int qemudDomainConnectNetBackend( unsigned int qemuCmdFlags) { qemuDomainObjPrivatePtr priv = vm->privateData; - char *tapfd_name = NULL; + char *hostnet_name = NULL; int tapfd = -1; char *netstr = NULL; int ret = -1; @@ -5664,27 +5673,30 @@ static int qemudDomainConnectNetBackend( if (tapfd < 0) return -1; - if (virAsprintf(&tapfd_name, "fd-%s", net->info.alias) < 0) { + if (virAsprintf(&hostnet_name, "host%s", net->info.alias) < 0) { virReportOOMError(conn); close(tapfd); return -1; } if (!(netstr = qemuBuildHostNetStr(conn, net, ' ', - vlan, tapfd_name))) { - VIR_FREE(tapfd_name); + vlan, hostnet_name))) { + VIR_FREE(hostnet_name); close(tapfd); return -1; } qemuDomainObjEnterMonitorWithDriver(driver, vm); - if (qemuMonitorSendFileHandle(priv->mon, tapfd_name, tapfd) < 0) + if (qemuMonitorRemoveHostNetwork(priv->mon, vlan, hostnet_name) < 0) + VIR_INFO0("Did not remove existing host network"); + + if (qemuMonitorSendFileHandle(priv->mon, hostnet_name, tapfd) < 0) goto out; if (qemuMonitorAddHostNetwork(priv->mon, netstr) < 0) { - if (qemuMonitorCloseFileHandle(priv->mon, tapfd_name) < 0) - VIR_WARN(_("Failed to close tapfd with '%s'"), tapfd_name); + if (qemuMonitorCloseFileHandle(priv->mon, hostnet_name) < 0) + VIR_WARN(_("Failed to close tapfd with '%s'"), hostnet_name); goto out; } @@ -5693,7 +5705,7 @@ static int qemudDomainConnectNetBackend( out: qemuDomainObjExitMonitorWithDriver(driver, vm); VIR_FREE(netstr); - VIR_FREE(tapfd_name); + VIR_FREE(hostnet_name); close(tapfd); return ret; } @@ -8549,6 +8561,72 @@ out: return ret; } +static void *qemudDomainReconnectNetBackends(void *opaque) +{ + virConnectPtr conn = virConnectOpen("qemu:///system"); + struct qemud_driver *driver = conn->privateData; + const unsigned char *uuid = opaque; + + while (1) { + virDomainObjPtr vm; + unsigned int qemuCmdFlags; + int i; + + usleep(1000000); + + VIR_ERROR(_("qemuDomainReconnectNetBackends (%p %p)"), + conn, driver); + + qemuDriverLock(driver); + vm = virDomainFindByUUID(&driver->domains, uuid); + if (!vm) { + VIR_ERROR0("qemuDomainReconnectNetBackends done"); + qemuDriverUnlock(driver); + break; + } + + VIR_ERROR(_("qemuDomainReconnectNetBackends (%s)"), + vm->def->name); + + if (qemudExtractVersionInfo(vm->def->emulator, + NULL, + &qemuCmdFlags) < 0) + goto cleanup; + + if (qemuDomainObjBeginJobWithDriver(driver, vm) < 0) + goto cleanup; + + if (!virDomainObjIsActive(vm)) + goto endjob; + + for (i = 0 ; i < vm->def->nnets ; i++) { + virDomainNetDefPtr net = vm->def->nets[i]; + + VIR_ERROR(_("qemuDomainReconnectNetBackends (%s)"), net->ifname); + + if (net->type != VIR_DOMAIN_NET_TYPE_DIRECT) + continue; + + if (qemudDomainConnectNetBackend(conn, driver, vm, + net, qemuCmdFlags) < 0) + VIR_ERROR(_("qemudDomainConnectNetBackend failed (%s)"), + net->ifname); + } + + endjob: + if (vm && + qemuDomainObjEndJob(vm) == 0) + vm = NULL; + + cleanup: + if (vm) + virDomainObjUnlock(vm); + qemuDriverUnlock(driver); + } + + return NULL; +} + static int qemuCPUCompare(virConnectPtr conn, const char *xmlDesc,

1 0