11:10:55 <crobinso> #startmeeting virt_test_day_f19
11:10:55 <zodbot> Meeting started Tue May 28 11:10:55 2013 UTC.  The chair is crobinso. Information about MeetBot at http://wiki.debian.org/MeetBot.
11:10:55 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
11:11:52 <kashyap> crobinso, Morning. would you like me to paste the scroll of previous chat ? So you might quickly glance through ? Or just let it be ?
11:11:53 <jclift_> kashyap: The 1.4.2-2 one is from fedora-updates-testing repo.
11:12:12 <jclift_> kashyap: Maybe the mirror it's using hasn't received the latest stuff yet.
11:12:14 <kashyap> jclift_, Hm, I previously pointed a build from koji, let me get that.
11:12:22 <kashyap> jclift_, Yes, that's a possibility.
11:12:23 <crobinso> kashyap: fpaste it please, Ill check it in a bit. thanks!
11:12:37 <kashyap> jclift_, There we go - http://koji.fedoraproject.org/koji/buildinfo?buildID=422227
11:12:40 <jclift_> kashyap: Actually, I have to get out of here in 40-odd minutes.
11:12:52 <kashyap> No worries, you can still do later too :)
11:12:59 <jclift_> It's 9:12pm atm, and the alarms go off in the office just after 10pm.
11:13:17 <jclift_> I should be able to pick this up in the morning though, if anyone's still around.
11:13:54 <kashyap> crobinso, http://www.fpaste.org/14942/ . No worries. You might just want to grep through "virt-manager" string :)
11:15:21 <crobinso> ignatenkobrain: the XML example in the serial login test was bogus, I fixed it now.
11:25:49 <ignatenkobrain> crobinso: thx. updated test result
11:29:25 <ignatenkobrain> QEMU: Checking for hardware virtualization                                 : WARN (Only emulated CPUs are available, performance will be significantly limited)
11:29:35 <ignatenkobrain> failed or passed ?
11:31:56 <crobinso> ignatenkobrain: hmm, something likely went wrong, but might be a small issue. is this nested virt?
11:32:06 <crobinso> ignatenkobrain: what's: dmesg | grep -i kvm
11:32:10 <ignatenkobrain> crobinso: yep
11:33:20 <ignatenkobrain> crobinso: http://paste.fedoraproject.org/14949/97407611 and in virt-manager ivy bridge not present. only sandy. I have ivy in host.
11:33:41 <jclift_> Interestingly, dmesg|grep -i kvm isn't showing any output
11:34:18 <crobinso> ignatenkobrain: what is: lsmod | grep kvm     and     ls -l /dev/kvm
11:34:57 <ignatenkobrain> crobinso: lsmod | grep kvm nothing show
11:35:59 <crobinso> ignatenkobrain: does this work: modprobe kvm_intel
11:36:04 <ignatenkobrain> crobinso: /dev/kvm no such file or directory
11:36:08 <crobinso> ignatenkobrain: and is qemu-system-x86 installed ?
11:37:06 <ignatenkobrain> crobinso: modprobe: ERROR: could not insert 'kvm_intel': Operation not supporte
11:37:52 <ignatenkobrain> crobinso: installed 1.4.2
11:38:43 <crobinso> ignatenkobrain: can you show dmesg output again, curious if that has any new messages after attempting the modprobe
11:39:15 <ignatenkobrain> crobinso: no hardware support
11:39:32 <ignatenkobrain> crobinso: my cpu have VT-x and VT-d
11:40:07 <crobinso> ignatenkobrain: can you paste the XML config of the guest: sudo virsh dumpxml <vmname>
11:41:34 <ignatenkobrain> crobinso: http://paste.fedoraproject.org/14956/41288136
11:43:55 <crobinso> ignatenkobrain: can you try this: shutdown the VM. sudo virsh edit <vmname>    remove the entire <cpu> block, and add <cpu model='host-passthrough'/>    start the guest and see if KVM works now
11:44:45 <crobinso> kashyap: any thoughts on the above? have you seen nested virt just 'not work' for intel ? I know it's still being worked on
11:47:41 <jclift_> Interesting.  "Unable to complete install: 'unsupported configuration: hda-duplex not supported in this QEMU binary"
11:48:02 <ignatenkobrain> crobinso: same.
11:49:08 <crobinso> jclift_: that has something to do with /dev/kvm permissions, I've been meaning to reproduce. try rmmod kvm && modprobe kvm, and then check ls -l /dev/kvm     it should have 666 perms
11:49:20 <jclift_> k
11:49:47 <ignatenkobrain> I will depart for an hour. crobinso, I write when I came back.
11:49:48 <crobinso> ignatenkobrain: yeah I guess nested virt just doesn't work for your intel host, nested virt for intel is still under development. please make a note of it in the test results and list your CPU model
11:49:59 <crobinso> ignatenkobrain: no problem, thanks for the thorough testing!
11:51:02 <jclift_> crobinso: crw-rw-rw-+ root kvm 10, 232 May 28 21:50 /dev/kvm
11:51:13 <jclift_> Both before and after the rmmod and mobprobe
11:51:59 <crobinso> jclift_: hmm. maybe restart libvirtd and see if it still reproduces that error. if not, a reboot fixed it for the previous reporter
11:53:27 <jclift_> Just restarted libvirtd.  Installer is functioning now.
11:53:41 <jclift_> Ugh, I have to split in abt 10 mins
11:57:16 <jclift_> Installer is working.  Installing the F19 Beta RC4 DVD into a vm.  One the banner during the installation where it says something like "I'm sorry Dave, I can't do that" it's all screwy artifacts.
11:57:50 <jclift_> s/One the/On the/
11:58:35 * jclift_ guesses it's something to do with the nouveau driver in the laptop
11:58:38 <jclift_> No real idea though
12:06:05 * jclift_ leaves for the night
12:14:03 <kashyap> crobinso, Sorry, stepped out for a walk. Reading scroll now
12:14:17 <crobinso> kashyap: no prob
12:16:52 <kashyap> ignatenkobrain, For /dev/kvm char device to reflect in your guest hypervisor,
12:17:07 <kashyap> Please run this on your host, $ virsh  capabilities | virsh cpu-baseline /dev/stdin
12:17:28 <kashyap> And paste the resultant output into the XML of your guest hypervisor
12:17:47 <kashyap> ignatenkobrain, It's explained here, section 3.2] - https://raw.github.com/kashyapc/nvmx-haswell/master/SETUP-nVMX.rst
12:18:07 <kashyap> And, another reference of guest hypervisor I posted here - https://github.com/kashyapc/nested-virt-notes-intel-f18/blob/master/libvirt-xmls/regular-guest-f18.xml
12:18:26 <kashyap> ignatenkobrain, Of note lines - 23 to 26
12:19:01 <kashyap> crobinso, Yes, I did see nested virt for Intel just 'not work' on more than one occasion
12:19:01 <crobinso> mrezanin: thanks for doing the test cases btw :)
12:19:16 <kashyap> It's still very fragile
12:19:18 <crobinso> kashyap: yeah i figure there's still some variability with hardware
12:19:33 <kashyap> However, significantly improved since F17 time frame.
12:19:55 <crobinso> kashyap: great work on exercising it though, I've seen your mails go upstream about testing the latest branches. it will be nice when this stuff 'just works'
12:20:13 <kashyap> crobinso, Yes, I've  been working with a couple of Intel folks
12:20:27 <kashyap> crobinso, Once nEPT patches are in mailine KVM,
12:20:35 <kashyap> There should be significant performance boost.
12:20:51 <kashyap> I've been swamped with some OpenStack work, so haven't worked on it this week at all :)
12:21:17 <kashyap> crobinso, And Gleb is diligently reviewing and providing comments.
12:25:15 <crobinso> kashyap: I still need to try the latest libvirt snapshot support, I know some bits I wanted before should be available now but haven't verified.
12:25:50 <kashyap> crobinso, Yes, did you have a chance to catch up with Christof Fergeau. He recently implemented something w/ snapshots.
12:25:56 <kashyap> Let me get a screenshot he passed over to me
12:26:31 <kashyap> crobinso, http://teuf.fedorapeople.org/virt-snapshot.png
12:26:53 <kashyap> I also still have to do further work w/ improvements in qemu block layer. But, yeah, only so much time :)
12:29:07 <crobinso> kashyap: hadn't seen that, but wanting something in virt-manager for a long time but just never had the chance to get to it. that and internal snapshot usage has been discouraged, and libvirt didn't provide all the support UI would want for external shapshots
12:31:31 <kashyap> crobinso, Yeah, probably you might just want to drop him a mail just to iron out some duplication work(if any). Indeed, I can see the state of affairs you're mentioning here. And true - nobody seriously cared to improve the perf. of internal snapshots.
12:31:53 <kashyap> Stefan Hajnoczi/Eric can give more insight there.
12:32:22 <crobinso> kashyap: apparently with qemu 1.5 stefan made it much faster, but anthony always discouraged their usage because they are non-live
12:32:31 <kashyap> crobinso, You're doing great work yourself trying to keep up with all the Upstream -> Fedora dance.
12:33:06 <kashyap> crobinso, Well, Anthony might discourage for his own reasons. But he cannot rule out the valid use-case of it.
12:33:21 <crobinso> kashyap: thanks. I feel like after a year I've finally got a hold of things and should start having freetime to devote to more bugfixing and features
12:33:39 <crobinso> kashyap: right, no doubt internal snapshots are still useful
12:33:49 <kashyap> Yes, I can sense the overwhelming effect of keeping up with bugs, upstream..etc..
12:34:11 <kashyap> I wish we could organize more frequently bug-triaging so that it doesn't bog you completely down with it.
13:08:03 <ignatenkobrain> crobinso: I am here. Now I continue testing
13:08:12 <crobinso> ignatenkobrain: great!
13:08:32 * rwmjones notes the libguestfs tests are here:
13:08:34 <rwmjones> http://libguestfs.org/guestfs-testing.1.html
13:08:40 <rwmjones> I'm running through them now on my machine
13:50:10 <rwmjones> yup, all the libguestfs tests work for me
15:07:02 * crobinso hears crickets
15:07:16 <crobinso> I've been filing a few bugs
15:07:18 <crobinso> https://bugzilla.redhat.com/show_bug.cgi?id=967927
15:07:31 <crobinso> .bug 967927
15:07:35 <zodbot> crobinso: Bug 967927 When deleting VM, hide storage list if 'delete storage' unselected - https://bugzilla.redhat.com/show_bug.cgi?id=967927
15:07:58 <crobinso> .bug 967914
15:08:03 <zodbot> crobinso: Bug 967914 ejecting cdrom in virt-manager gives 'tray is locked' error, but succeeds on next try - https://bugzilla.redhat.com/show_bug.cgi?id=967914
15:14:40 * rwmjones wonders if double frees are exploitable?
15:14:55 <rwmjones> because coverity found one in libguestfs in F19
15:15:02 <rwmjones> $ qemu-img create -f qcow2 -b /dev/fedora/f20rawhidex64 /tmp/test.qcow2
15:15:10 <rwmjones> $ guestfish -i -a /tmp/test.qcow2 -- rm /etc/redhat-release : touch /etc/redhat-release
15:15:22 <rwmjones> $ virt-inspector /tmp/test.qcow2
15:15:22 <rwmjones> *** glibc detected *** virt-inspector: double free or corruption (fasttop): 0x00007f18bc9925a0 ***
15:15:25 <rwmjones> ======= Backtrace: =========
15:15:27 <rwmjones> /lib64/libc.so.6(+0x34ecc7ca8e)[0x7f18b8e64a8e]
15:15:30 <rwmjones> /lib64/libguestfs.so.0(+0x3f91898078)[0x7f18ba13c078]
15:15:32 <rwmjones> /lib64/libguestfs.so.0(+0x3f91899761)[0x7f18ba13d761]
15:15:35 <rwmjones> /lib64/libguestfs.so.0(+0x3f91896d12)[0x7f18ba13ad12]
15:15:37 <rwmjones> /lib64/libguestfs.so.0(+0x3f91894140)[0x7f18ba138140]
15:15:40 <rwmjones> /lib64/libguestfs.so.0(guestfs_inspect_os+0x35)[0x7f18ba0bcc35]
15:15:42 <rwmjones> virt-inspector(main+0x547)[0x7f18ba7c57d7]
15:15:45 <rwmjones> /lib64/libc.so.6(__libc_start_main+0xf5)[0x7f18b8e09a05]
15:15:47 <rwmjones> virt-inspector(+0x6665)[0x7f18ba7c7665]
16:13:23 <crobinso> I added a new standard test case for CDROM media change: https://fedoraproject.org/wiki/QA:Testcase_Virt_Change_CDROM_Media
16:13:42 <crobinso> kinda standard functionality that regresses a surprising amount :)
16:23:39 <ignatenkobrain> crobinso: https://fedoraproject.org/wiki/QA:Testcase_Virtualization_URL_Guest_Install update this
16:24:07 <ignatenkobrain> crobinso: and add this bug to known issuses: https://bugzilla.redhat.com/show_bug.cgi?id=967780
16:30:41 <crobinso> ignatenkobrain: will do
16:32:20 <ignatenkobrain> crobinso: thanks
19:40:38 <crobinso> unfortunately the storage migration case is busted, libvirt apparently doesn't work with --tunnelled --copy-storage-all option pair, but it wasn't explicitly erroring
19:40:41 <crobinso> bug coming in a bit
19:40:50 <crobinso> trying to get the test case actually working as well
20:48:33 <knoel> eblake, ping?
20:51:25 <eblake> pong
21:01:02 <eblake> knoel: pong
21:01:30 <knoel> eblake, I can't make virsh iface-bridge work
21:01:56 <eblake> sounds more like something laine is familiar with, but I can try to help as well
21:02:12 <knoel> eblake, laine, it says: "error: An error occurred, but the cause is unknown"
21:02:17 <eblake> are you using the testday live image, an F19 install, or something else?
21:02:50 <knoel> eblake, laine, F19 Beta on DVD
21:03:29 <knoel> 1.0.5.1
21:06:04 <eblake> do any of the logs contain more details?  For example, check /var/log/libvirt/libvirtd.log
22:24:21 <eblake> wow - I just noticed that /var/log/libvirt/libvirtd.log doesn't exist on my F18+fedora-virt-preview setup; that smells like a regression compared to what older libvirt did
23:11:33 <rydhwan> how can install live usb sugar in hard disk merci
23:21:36 <satellit_e> go to terminal in sugar (left top bar list view of f3 Home) go su to root and type "liveinst"
23:21:53 <satellit_e> right corner of bar*
23:22:11 <satellit_e> terminal is at the bottom of list of applications
00:11:07 <crobinso> #endmeeting