August 2010 - Devel - libvirt List Archives

[libvirt] [PATCH v2] Support for qemu aio drive option
by Matthias Dahl 05 Jan '11

05 Jan '11

Revised patch against libvirt 0.7.6 to support qemu's aio option. qemu allows the user to choose what io storage api should be used, either the default (threads) or native (linux aio) which in the latter case can result in better performance. This patch exposes this functionality through libvirt. Thanks a lot to Eric Blake and Matthias Bolte for their comments. --- docs/formatdomain.html.in | 4 +++- docs/schemas/domain.rng | 13 +++++++++++-- src/conf/domain_conf.c | 23 +++++++++++++++++++++++ src/conf/domain_conf.h | 10 ++++++++++ src/qemu/qemu_conf.c | 14 ++++++++++++++ src/qemu/qemu_conf.h | 2 ++ 6 files changed, 63 insertions(+), 3 deletions(-) diff --git a/docs/formatdomain.html.in b/docs/formatdomain.html.in index ce49f7d..ec85ef3 100644 --- a/docs/formatdomain.html.in +++ b/docs/formatdomain.html.in @@ -478,7 +478,9 @@ attribute is the primary backend driver name, while the optional <code>type</code> attribute provides the sub-type. The optional <code>cache</code> attribute controls the cache mechanism, possible values are "default", "none", - "writethrough" and "writeback". <span class="since">Since 0.1.8</span> + "writethrough" and "writeback". Specific to KVM guests is the optional <code>aio</code> + attribute which controls what storage api is used for io operations, its possible + values are "threads" and "native". <span class="since">Since 0.1.8; "aio" attribute since 0.7.7</span> </dd> <dt><code>encryption</code></dt> <dd>If present, specifies how the volume is encrypted. See diff --git a/docs/schemas/domain.rng b/docs/schemas/domain.rng index bb6d00d..5d2dafe 100644 --- a/docs/schemas/domain.rng +++ b/docs/schemas/domain.rng @@ -499,8 +499,9 @@ <ref name="driverCache"/> </group> </choice> - <empty/> - </element> + <optional> + <ref name="driverAIO"/> + </optional> </define> <define name="driverFormat"> <attribute name="name"> @@ -521,6 +522,14 @@ </choice> </attribute> </define> + <define name="driverAIO"> + <attribute name="aio"> + <choice> + <value>threads</value> + <value>native</value> + </choice> + </attribute> + </define> <define name="controller"> <element name="controller"> <optional> diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index 766993c..8bee7b8 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -120,6 +120,11 @@ VIR_ENUM_IMPL(virDomainDiskCache, VIR_DOMAIN_DISK_CACHE_LAST, "writethrough", "writeback") +VIR_ENUM_IMPL(virDomainDiskAIO, VIR_DOMAIN_DISK_AIO_LAST, + "default", + "native", + "threads") + VIR_ENUM_IMPL(virDomainController, VIR_DOMAIN_CONTROLLER_TYPE_LAST, "ide", "fdc", @@ -1228,6 +1233,7 @@ virDomainDiskDefParseXML(virConnectPtr conn, char *target = NULL; char *bus = NULL; char *cachetag = NULL; + char *aiotag = NULL; char *devaddr = NULL; virStorageEncryptionPtr encryption = NULL; char *serial = NULL; @@ -1293,6 +1299,7 @@ virDomainDiskDefParseXML(virConnectPtr conn, driverName = virXMLPropString(cur, "name"); driverType = virXMLPropString(cur, "type"); cachetag = virXMLPropString(cur, "cache"); + aiotag = virXMLPropString(cur, "aio"); } else if (xmlStrEqual(cur->name, BAD_CAST "readonly")) { def->readonly = 1; } else if (xmlStrEqual(cur->name, BAD_CAST "shareable")) { @@ -1409,6 +1416,13 @@ virDomainDiskDefParseXML(virConnectPtr conn, goto error; } + if (aiotag && + (def->aiomode = virDomainDiskAIOTypeFromString(aiotag)) < 0) { + virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, + _("unknown disk aio mode '%s'"), aiotag); + goto error; + } + if (devaddr) { if (sscanf(devaddr, "%x:%x:%x", &def->info.addr.pci.domain, @@ -1450,6 +1464,7 @@ cleanup: VIR_FREE(driverType); VIR_FREE(driverName); VIR_FREE(cachetag); + VIR_FREE(aiotag); VIR_FREE(devaddr); VIR_FREE(serial); virStorageEncryptionFree(encryption); @@ -4565,6 +4580,7 @@ virDomainDiskDefFormat(virConnectPtr conn, const char *device = virDomainDiskDeviceTypeToString(def->device); const char *bus = virDomainDiskBusTypeToString(def->bus); const char *cachemode = virDomainDiskCacheTypeToString(def->cachemode); + const char *aiomode = virDomainDiskAIOTypeToString(def->aiomode); if (!type) { virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, @@ -4586,6 +4602,11 @@ virDomainDiskDefFormat(virConnectPtr conn, _("unexpected disk cache mode %d"), def->cachemode); return -1; } + if (!aiomode) { + virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, + _("unexpected disk aio mode %d"), def->aiomode); + return -1; + } virBufferVSprintf(buf, " <disk type='%s' device='%s'>\n", @@ -4597,6 +4618,8 @@ virDomainDiskDefFormat(virConnectPtr conn, virBufferVSprintf(buf, " type='%s'", def->driverType); if (def->cachemode) virBufferVSprintf(buf, " cache='%s'", cachemode); + if (def->aiomode) + virBufferVSprintf(buf, " aio='%s'", aiomode); virBufferVSprintf(buf, "/>\n"); } diff --git a/src/conf/domain_conf.h b/src/conf/domain_conf.h index 0b79e88..07805a6 100644 --- a/src/conf/domain_conf.h +++ b/src/conf/domain_conf.h @@ -141,6 +141,14 @@ enum virDomainDiskCache { VIR_DOMAIN_DISK_CACHE_LAST }; +enum virDomainDiskAIO { + VIR_DOMAIN_DISK_AIO_DEFAULT, + VIR_DOMAIN_DISK_AIO_NATIVE, + VIR_DOMAIN_DISK_AIO_THREADS, + + VIR_DOMAIN_DISK_AIO_LAST +}; + /* Stores the virtual disk configuration */ typedef struct _virDomainDiskDef virDomainDiskDef; typedef virDomainDiskDef *virDomainDiskDefPtr; @@ -154,6 +162,7 @@ struct _virDomainDiskDef { char *driverType; char *serial; int cachemode; + int aiomode; unsigned int readonly : 1; unsigned int shared : 1; virDomainDeviceInfo info; @@ -897,6 +906,7 @@ VIR_ENUM_DECL(virDomainDisk) VIR_ENUM_DECL(virDomainDiskDevice) VIR_ENUM_DECL(virDomainDiskBus) VIR_ENUM_DECL(virDomainDiskCache) +VIR_ENUM_DECL(virDomainDiskAIO) VIR_ENUM_DECL(virDomainController) VIR_ENUM_DECL(virDomainFS) VIR_ENUM_DECL(virDomainNet) diff --git a/src/qemu/qemu_conf.c b/src/qemu/qemu_conf.c index 3d83a8f..fd3a670 100644 --- a/src/qemu/qemu_conf.c +++ b/src/qemu/qemu_conf.c @@ -83,6 +83,12 @@ VIR_ENUM_IMPL(qemuDiskCacheV2, VIR_DOMAIN_DISK_CACHE_LAST, "writethrough", "writeback"); +VIR_ENUM_DECL(qemuDiskAIO) +VIR_ENUM_IMPL(qemuDiskAIO, VIR_DOMAIN_DISK_AIO_LAST, + "default", + "native", + "threads"); + VIR_ENUM_DECL(qemuVideo) VIR_ENUM_IMPL(qemuVideo, VIR_DOMAIN_VIDEO_TYPE_LAST, @@ -1137,6 +1143,8 @@ static unsigned int qemudComputeCmdFlags(const char *help, flags |= QEMUD_CMD_FLAG_DRIVE_CACHE_V2; if (strstr(help, "format=")) flags |= QEMUD_CMD_FLAG_DRIVE_FORMAT; + if (strstr(help, "aio=threads|native")) + flags |= QEMUD_CMD_FLAG_DRIVE_AIO; } if (strstr(help, "-vga") && !strstr(help, "-std-vga")) flags |= QEMUD_CMD_FLAG_VGA; @@ -2340,6 +2348,12 @@ qemuBuildDriveStr(virDomainDiskDefPtr disk, virBufferAddLit(&opt, ",cache=off"); } + if (disk->aiomode && (qemuCmdFlags & QEMUD_CMD_FLAG_DRIVE_AIO)) { + const char * mode = qemuDiskAIOTypeToString(disk->aiomode); + + virBufferVSprintf(&opt, ",aio=%s", mode); + } + if (virBufferError(&opt)) { virReportOOMError(NULL); goto error; diff --git a/src/qemu/qemu_conf.h b/src/qemu/qemu_conf.h index 101f187..780ae6e 100644 --- a/src/qemu/qemu_conf.h +++ b/src/qemu/qemu_conf.h @@ -82,6 +82,8 @@ enum qemud_cmd_flags { QEMUD_CMD_FLAG_SDL = (1 << 27), /* Is the new -sdl arg available */ QEMUD_CMD_FLAG_SMP_TOPOLOGY = (1 << 28), /* Is sockets=s,cores=c,threads=t available for -smp? */ QEMUD_CMD_FLAG_NETDEV = (1 << 29), /* The -netdev flag & netdev_add/remove monitor commands */ + + QEMUD_CMD_FLAG_DRIVE_AIO = (1 << 30), /* Is -drive aio= avail */ }; /* Main driver state */ -- 1.7.0.4

4 14

[libvirt] ruby-libvirt issue
by Jaromír Červenka 25 Nov '10

25 Nov '10

Good evening, could anybody explain me this, please? divinus:~ # virsh error: cannot recv data: : Connection reset by peer error: failed to connect to the hypervisor divinus:~ # I am trying to imlement Chris's ruby-libvirt to my Ruby on Rails application. First few connection to libvirt daemon (thru ruby-libvirt API) from my app are ok, but then somethings happend and it's not possible to connect anymore until I restart libvirt daemon manually by init.d script or until I restart apache (or WEBrick). And it's not possible connect to virsh from shell also! See messages above. When I try ruby-libvirt from bare Ruby script, everythig is fine - doesn't matter how many times I connect. It's strange. Does anybody have experience with ruby-libvirt in RoR application? Thank you. Jaromir.

6 11

[libvirt] VMware Workstation/Player support
by Jean-Baptiste Rouault 18 Nov '10

18 Nov '10

Hi all, I'm working on the hynesim project (http://www.hynesim.org http://blog.hynesim.org) We are going to add VMware Workstation support to hynesim, so we would like to contribute to the development of a libvirt driver. There probably are multiple ways of doing this, one of them could be to use the VIX C API provided by VMware : http://www.vmware.com/support/developer/vix-api/ However, the VIX license seems to be quite restrictive : http://www.vmware.com/download/eula/vixapi_eula.html I'm not a license expert so I don't know if this license forbids using the API in software like libvirt. Another possibility could be to run VMware command line tools from libvirt to control guests. As Daniel P. Berrange stated on IRC yesterday, the vmx <-> XML conversion code can be reused from the ESX driver. I'd like to have your opinion about the "best" to implement this VMware Workstation driver. Regards, Jean-Baptiste Rouault

4 7

[libvirt] process= support for 'qemu-kvm -name' [Bug 576950]
by John Morrissey 17 Nov '10

17 Nov '10

I wrote (attached here, and to the bug) a quick patch that sets the process name to the same value as the window title. I'm unsure where to go from here. Should I add support for converting "native" QEMU command lines to libvirt XML? What would that look like, since I'm not modifying the libvirt format? Should it just drop any ,process= from the QEMU command line it's parsing? I also imagine the test cases will need updating. If someone could give me some high-level guidance, I'd be happy to keep going on this. john -- John Morrissey _o /\ ---- __o jwm(a)horde.net _-< \_ / \ ---- < \, www.horde.net/ __(_)/_(_)________/ \_______(_) /_(_)__

4 11

[libvirt] New save/restore api proposal
by Jean-Baptiste Rouault 13 Nov '10

13 Nov '10

Hello all, I'd like to add support for save and restore to the OpenVZ and VirtualBox drivers because I have to support these operations in the application I'm working on. However, the save/restore API in its current state doesn't fit well to our needs. The main problem is that the domain definition is included inside the save file. This is problematic because between the save and the restore operations, the names of the network interfaces on the host side are likely to have changed and we can't modify them before restoring the domain. To summarize, what we would like to be able to do is: - save a domain and undefine it - update the domain definition to use the new names of host side interfaces - restore the domain This is why I would like to add two new functions with the following signatures (better names are probably needed): int virDomainSaveState(virDomainPtr domain, const char *to); int virDomainRestoreState(virDomainPtr domain, const char *from); The first one would do the same as virDomainSave but without prepending the domain's definition to the save file. The other function would be able to restore a domain saved by the first one. What do you think ? Regards, Jean-Baptiste

2 3

[libvirt] [PATCH 0/2] speed up qemu domain save by increasing dd blocksize
by Laine Stump 29 Oct '10

29 Oct '10

These two patches are posted together because applying the 2nd exposes the bug fixed by the first. Here are the results of tests I made with various block sizes before deciding the 1MB really was the best balance (all tests were done on a paused 512MB domain, saving to local disk on a Lenovo T61 laptop) BS M:SS save image size ----- ---- --------------- 2048K - 0:56 476135451 1024K - 0:56 475090953 512k - 1:02 474564173 256k - 1:10 474303797 128k - 1:25 474176859 512 - 3:47 474085423 - the original I didn't bother testing sizes between 512 and 128k, as there was still significant improvement from 128k to 256k.

5 27

[libvirt] web page suggestion
by Eric Blake 23 Oct '10

23 Oct '10

My XSL skills are less than stellar, so I'm throwing this out to the list in case someone else can pick it up and come up with a decent patch in less time. Right now, http://libvirt.org/ChangeLog.html is worthless; it is linked from a couple of other pages, such as http://libvirt.org/news.html. A better place to link would be a live git page: http://libvirt.org/git/?p=libvirt.git;a=log I don't know whether it is easier to update news.html.in and sitemap.html.in to point directly to the new link, or if we should keep ChangeLog.xsl but have it revamped to point to the new link. -- Eric Blake eblake(a)redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org

3 4

[libvirt] KVM Forum 2010: presentations online
by Chris Wright 20 Oct '10

20 Oct '10

KVM Forum 2010 was quite a success, many thanks to all who participated! For those who couldn't attend, the presentations are available online now: (thanks to Andrew Cathrow for pushing them all up) http://www.linux-kvm.org/page/KVM_Forum_2010#Presentations We were also able to video the speakers, and will send a note when the videos are available. (and thanks again to Andrew Cathrow for making this happen) thanks, -chris

1 1

[libvirt] Xen string2sexpr and sexpr2string lose quotes?
by Thomas Graves 14 Oct '10

14 Oct '10

Hello all, I am running xen on rhel5 and using libvirt0.7.2 (I also tried 0.7.7) and it looks like the routines string2sexpr and sexpr2string seem to lose the quotes around the image args in the configuration. Has anyone seen this and have a patch for this? I have the following libvirt config: <os> <type>linux</type> <kernel>/usr/lib/xen/boot/pv-grub-x86_64.gz</kernel> <cmdline>(hd0,0)/grub/menu.lst</cmdline> </os> It generates the xm config info: (image (linux (kernel /usr/lib/xen/boot/pv-grub-x86_64.gz) (args '(hd0,0)/grub/menu.lst') (device_model /usr/lib64/xen/bin/qemu-dm) ) ) I call virDomainSetAutostart on the domain and traced it through and saw that it gets the string quoted (args '(hd0,0)/grub/menu.lst') from xen then ends up calling string2sexpr, changes the xend_on_start, and then sexpr2string, and it ends up without quotes (args (hd0,0)/grub/menu.lst) and that is what it sends back to xen. Xen then seems to chop it off to (args ('hd0,0')) Thanks, Tom

3 2

[libvirt] [PATCH] document authentication with PolicyKit-1
by Gerd v. Egidy 07 Oct '10

07 Oct '10

Hi, I just encountered the same problem with the newer PolicyKit as discussed e.g. in this thread: http://lists.fedoraproject.org/pipermail/virt/2010-June/002081.html I solved it and thought updating the documentation for PolicyKit 1 would be nice. Patch attached. Kind regards, Gerd -- Address (better: trap) for people I really don't want to get mail from: jonas(a)cactusamerica.com

4 4