Re: [Users] about live snapshot and qemu-kvm

On Mon, 2014-02-17 at 09:59 -0500, Douglas Schilling Landgraf wrote: > On 02/17/2014 02:00 AM, Karli Sjöberg wrote: >> On Thu, 2014-02-13 at 14:42 -0500, Douglas Schilling Landgraf wrote: >>> On 01/01/2014 06:24 AM, Gianluca Cecchi wrote: >>>> On Wed, Jan 1, 2014 at 4:38 AM, R P Herrold wrote: >>>> >>>>> >>>>> Out of curiousity, _what_ build environment 'flags' do you >>>>> all, participating in this thread, refer to? -- the thread >>>>> does not enumerate them explicitly, and one cannot expect to >>>>> hit by 'indirect fire', a target not exposed >>>>> >>>>> With best regards, this New Year's eve >>>> >>>> I'm far from being a programmer, but as I went to compare build >>>> environments, between >>>> qemu-kvm-rhev-0.12.1.2-2.415.el6_5.3.src.rpm >>>> and >>>> qemu-kvm-0.12.1.2-2.415.el6_5.3.src.rpm >>>> in related spec file I see >>>> [g.cecchi@tekkaman SPECS]$ diff qemu-kvm.spec.upstream qemu-kvm.spec.rhev >>>> 3c3 >>>> < %define rhev 0 >>>> --- >>>>> %define rhev 1 >>>> 12928a12929 >>>>>>>>>>>> rhel-6.5 >>>> >>>> and apart other probably not trivial implications, such as guest agent >>>> part, I see that the "configure" command takes one extra argument in >>>> base RH EL 6.5, that is >>>> >>>> --disable-rhev-features >>>> >>>> The only patch file containing this keyword is >>>> >>>> kvm-Block-streaming-disable-for-RHEL.patch >>>> >>>> and inside it there are these lines that impacts configure options and >>>> related built qemu-kvm: >>>> >>>> --- a/configure >>>> +++ b/configure >>>> @@ -286,6 +286,7 @@ spice="" >>>> smartcard="" >>>> smartcard_nss="" >>>> live_snapshots="yes" >>>> +block_stream="yes" >>>> usb_redir="" >>>> >>>> # OS specific >>>> @@ -686,10 +687,22 @@ for opt do >>>> ;; >>>> --enable-live-snapshots) live_snapshots="yes" >>>> ;; >>>> + --disable-block-stream) block_stream="no" >>>> + ;; >>>> + --enable-block-stream) block_stream="yes" >>>> + ;; >>>> --disable-usb-redir) usb_redir="no" >>>> ;; >>>> --enable-usb-redir) usb_redir="yes" >>>> ;; >>>> + --disable-rhev-features) >>>> + live_snapshots="no"; >>>> + block_stream="no"; >>>> + ;; >>>> + --enable-rhev-features) >>>> + live_snapshots="yes"; >>>> + block_stream="yes"; >>>> + ;; >>>> *) echo "ERROR: unknown option $opt"; show_help="yes" >>>> ;; >>>> esac >>>> @@ -863,8 +876,12 @@ echo " --disable-smartcard-nss disable >>>> smartcard nss support" >>>> echo " --enable-smartcard-nss enable smartcard nss support" >>>> echo " --disable-live-snapshots disable live block device snapshot support" >>>> echo " --enable-live-snapshots enable live block device snapshot support" >>>> +echo " --disable-block-stream disable block streaming support" >>>> +echo " --enable-block-stream enable block streaming support" >>>> echo " --disable-usb-redir disable usb network redirection support" >>>> echo " --enable-usb-redir enable usb network redirection support" >>>> +echo " --disable-rhev-features disable RHEV-only features" >>>> +echo " --enable-rhev-features enable RHEV-only features" >>>> echo "" >>>> echo "NOTE: The object files are built at the place where configure >>>> is launched" >>>> exit 1 >>>> @@ -2271,6 +2288,7 @@ echo "Trace backend $trace_backend" >>>> echo "spice support $spice" >>>> echo "nss used $smartcard_nss" >>>> echo "Live snapshots $live_snapshots" >>>> +echo "Block streaming $block_stream" >>>> echo "xfsctl support $xfs" >>>> echo "usb net redir $usb_redir" >>>> >>>> @@ -2526,6 +2544,10 @@ if test "$live_snapshots" = "yes" ; then >>>> echo "CONFIG_LIVE_SNAPSHOTS=y" >> $config_host_mak >>>> fi >>>> >>>> +if test "$block_stream" = "yes" ; then >>>> + echo "CONFIG_BLOCK_STREAM=y" >> $config_host_mak >>>> +fi >>>> + >>>> if test "$usb_redir" = "yes" ; then >>>> echo "CONFIG_USB_REDIR=y" >> $config_host_mak >>>> fi >>>> >>>> I don't think the rhev argument has instead implications in upstream >>>> source qemu-kvm-0.12.1.2.tar.gz >>>> So I think that if you want to dig more and if you have more >>>> competences, you have to see the full spec file and the full patch >>>> above. >>>> >>>> Files downloaded here: >>>> >>>> upstream >>>> http://ftp.redhat.com/redhat/linux/enterprise/6Server/en/os/SRPMS/qemu-kv... >>>> >>>> rhev >>>> http://ftp.redhat.com/redhat/linux/enterprise/6Server/en/RHEV/SRPMS/qemu-... >>>> >>> >>> Just for the record, we have setup a jenkins job to rebuild qemu-kvm for >>> el6 until we get it officially from centos: >>> http://jenkins.ovirt.org/view/Packaging/job/qemu-kvm-rhev_create_rpms_el6/ >>> >>> -- >>> Cheers >>> Douglas >>> _______________________________________________ >>> Users mailing list >>> Users(a)ovirt.org >>> http://lists.ovirt.org/mailman/listinfo/users >> >> I think this calls for an oldschool question/statement; >> >> Who da man? ... You da man! :) >> >> Live snapshot now just works. Haven´t verified quiesce yet though, but >> looks good as far as the events in webadmin is concerned. Next to test >> is live disk migration. >> >> > Thanks a lot for the feedback :) > > -- > Cheers > Douglas Hey, have gone through the logs to verify quiescence and the only thing actually mentioning it comes from vdsm.log: Thread-127371::DEBUG::2014-02-19 14:54:15,612::BindingXMLRPC::977::vds::(wrapper) return vmSnapshot with {'status': {'message': 'Done', 'code': 0}, 'quiesce': False} What is it saying really? That quiescing failed, or that it was deliberately taken without it, or what? Nothing in qemu-ga.log either, even if changed to "--verbose". Like it never got any calls to quiesce in the first place. The guest does have the device[1] attached at least. The guest is CentOS-6.4 or 5, if that matters. [1]: /dev/virtio-ports/org.qemu.guest_agent.0