October 2009 - virt - Fedora mailing-lists

libvirtd crashes host when autostarting
by Rich Mahn 06 Jan '10

06 Jan '10

Summary: My host crashes when I use the autostart function in libvirtd. This command sequence works fine: service libvirtd stop virsh autostart mydomain --disable service lib virtd start virsh start mydomain This command sequence will crash the host: service libvirtd stop virsh autostart mydomain service libvirtd start I can't find any error messages related to this. The host system freezes, the keyboard starts blinking the caps-lock and scroll-lock, and as best I can tell, the kernel has stopped running. If I starting up the host system with the VM autostarted, I sometimes get messages that look like kernel errors on the screen, but I don't find them in any log files so far. They seem to be related to networking. Here's some configuration information--I don't know really what all is needed, but this, I guess is a start: Host and VM both Fedora 11 with the lastest updates. Host machine in Intel based with the virtual support turned on. In the /var/log/libvirtd/qemu directory, the command line to start the virtual machine is different when started from autostart. The difference is in this portion: -net tap,fd=11,vlan=0 for failure (autostart) and -net tap,fd=18,vlan=0 for success I am using a bridged connection. Here's the command line that's used, split over lines for some degree of readability: LC_ALL=C PATH=/sbin:/usr/sbin:/bin:/usr/bin /usr/bin/qemu-kvm -S -M pc -m 1024 -smp 1 -name mydomain -uuid 5cde5a23-2abe-fcdd-04dc-5ae3a46ab51c -monitor pty -pidfile /var/run/libvirt/qemu//mydomain.pid -boot c -drive file=,if=ide,media=cdrom,index=2 -drive file=/dev/mapper/vg_host-lv_mydomain,if=virtio,index=0,boot=on -drive file=/dev/mapper/vg_host-lv_mydomain_data,if=virtio,index=1 -net nic,macaddr=54:52:00:67:91:e9,vlan=0,model=virtio -net tap,fd=11,vlan=0 -serial pty -parallel none -usb -usbdevice tablet -vnc 127.0.0.1:0

6 9

disabling ksm by default
by Izik Eidus 11 Nov '09

11 Nov '09

Hi, I saw that Fedora 12 will have ksm tunning script that control the ksm speed / kernel pages allocation. The only problem that I have is - as far as I remember ksm is by default enabled and only with 2000 kernel pages... What I am worried about is that users wont use the ksm tunning script and would just run ksm with this 2000 kernel pages - the result would be that ksm will probably merge just the zero pages (that could be alot of memory) and the user might not know that much more memory can be saved... Is it possible to at least make ksm disabled by default? so the users will have to run the ksm tunning script when they want to start ksm? And if we set it to disabled by default, cant we set the initialized values into more realistic value? (like 1/4 of the memory in the mainline kernels? ) Thanks.

8 19

duplicated console and details buttons
by Gianluca Cecchi 30 Oct '09

30 Oct '09

You get the "console" and "details" buttons/functions both inside the toolbar and as labels just under the toolbar. It is ok that if you uncheck "view toolbar" you still get the labels, but it seems to me sort of redundant... they are right one below the same... see attached image Just my opinion. Gianluca

2 4

Fedora virt status
by Mark McLoughlin 30 Oct '09

30 Oct '09

Fedora 12 ========= Fedora 12 Beta slipped by a week, but was released on Oct 20: https://fedoraproject.org/wiki/F12_Beta_Announcement We're now in the final countdown to F12 GA. The first release candidate will be composed next week and, all going well, we should have a release on November 17. Thankfully, the virt blocker list is now clear, but if you're looking to help with making Fedora 12 even better, there's no better place to start than the F12 target tracker bug: https://bugzilla.redhat.com/showdependencytree.cgi?id=F12VirtTarget&hide_re… There's over 100 bugs there that need your help! Witty Tagline ============= Dan Berrange has done a great job of coming up with a "witty tagline" for each of the virtualization feature sets in previous Fedora releases: https://fedoraproject.org/wiki/Virtualization/History Fedora 11: The walled garden Fedora 10: Management at a distance Fedora 9: Farewell to old friends Fedora 8: Protection from the bad guys Fedora 7: The new kid on the block Fedora Core 6: Virtualization grows up Fedora Core 5: The future is now Fedora Core 4: Glimpse of the future We need a tagline which describes our eclectic mix of features in Fedora 12. Any ideas? Fedora Weekly News ================== Dale Bewley has contributed another bunch of virtualization sections to Fedora Weekly News: https://fedoraproject.org/wiki/FWN/Issue196#Virtualization https://fedoraproject.org/wiki/FWN/Issue197#Virtualization https://fedoraproject.org/wiki/FWN/Issue199#Virtualization ABRT and qemu-kvm ================= It's quite common that we have to ask bug reports to collect a stack trace for a qemu-kvm segfault. Is ABRT the future? https://bugzilla.redhat.com/show_bug.cgi?id=528754#c1 Try this: $> yum install --enablerepo=rawhide-debuginfo qemu-debuginfo $> yum install -y abrt abrt-cli abrt-addon-ccpp abrt-plugin-logger $> service abrtd restart then when qemu-kvm crashes, you should be able to do e.g.: $> abrt-cli --get-list $> echo n | abrt-cli --report $uuid > t.log Anyone got ideas for a better way to use ABRT to catch virt related bugs? We could do with adding instructions to: http://fedoraproject.org/wiki/How_to_debug_Virtualization_problems Got Windows Guests? =================== Rich Jones has added yet another weird and fantastical feature to libguestfs ... and he needs your help to check that it works: http://www.redhat.com/archives/fedora-virt/2009-October/msg00128.html If you have any Windows guests, then you can help Fedora to support Windows guests better by spending a few minutes testing the Windows Registry feature we just added to libguestfs 1.0.75. Basically, the new virt-win-reg tool can be used to display entries from the registry in a Windows guest. How cool is that? Bugzilla ======== DOOM-O-METER: 198 bugs open four weeks ago, only 186 now! Ongoing Bugs ============ == kernel == https://bugzilla.redhat.com/515741 2.6.30 kernel stopped supporting xattrs on hugetlbfs John Cooper is working on getting a patch together for F-12. https://bugzilla.redhat.com/524808 swiotlb should be enabled when VT-d setup fails dwmw2 points out that if VT-d setup fails (because e.g. a bad BIOS is detected), then we should make sure that swiotlb is still enabled. Chris Wright has been working on the fix. All are agreed that this shouldn't be considered a release blocker. https://bugzilla.redhat.com/530389 allow userspace to adjust kvmclock offset Glauber has patches queued upstream to add new kvmclock ioctls and wants them backported to Fedora 12. https://bugzilla.redhat.com/531270 TG3, kvm, ipv6 & tso data corruption bug Rik van Riel discovered this TSO6 related issue with the tg3 driver which is causing traffic from his KVM guest to be corrupted. https://bugzilla.redhat.com/528245 starting a KVM guest before wireless NIC is brought up results in "nf_conntrack: table full, dropping packet." Paul Lambert reports that his machine locks up on boot if he sets his KVM guest to autostart. It looks like it may be related to the wireless connection not being available at that point, but we're still pretty far from narrowing the issue down. == qemu == https://bugzilla.redhat.com/528754 qemu-kvm segfault caused by -soundhw es1370 Gene Czarcinski reported this qemu segfault and managed to obtain a stack trace for it. https://bugzilla.redhat.com/528134 qemu should set O_CLOEXEC on all file descriptors glibc is now forking a ptchown program under certain circumstances, and SELinux issues showed that we are leaking file descriptors into forked processes. Uli claims we should set O_CLOEXEC but Dan, Markus and Kevin disagree. https://bugzilla.redhat.com/472236 KVM guests cannot boot PXE "local" This bug was originally reported against Fedora 10, but apparently is still a problem in Fedora 12 with gPXE. https://bugzilla.redhat.com/528764 qemu's SDL backend doesn't release the console framebuffer on exit Matt Booth discovered this annoyance with qemu's SDL backend. == libvirt == https://bugzilla.redhat.com/527736 libvirt can't create storage pool volumes on a FAT32 hard disk - 'cannot set file owner ... Operation not permitted' Reported by Richard Hughes, looks like libvirt's fs storage backend just isn't handling this case correctly. https://bugzilla.redhat.com/529363 virsh restore fails - failing to re-label the save file ? sVirt related virsh restore failure. == virtinst == https://bugzilla.redhat.com/528947 virtinst: Creating LVM volume on QEMU during VM creation fails A virtinst traceback during cloning, fixed by Cole upstream but needing backporting. == virt-manager == https://bugzilla.redhat.com/530210 virt-manager setfacl() fails on filesystems not mounted with the 'acl' option Our solution for "qemu can't access my homedir" was to use setfacl to give it access. We're scuppered in the case where users have created the filesystem themselves without the 'acl' option listed in the defaults. We need a better error message for this case. https://bugzilla.redhat.com/530948 virt-manager: switching VM to fullscreen pins the window to 1st virtual desktop Stefan Assman is seeing very strange behaviour when trying to use virt-manager in fullscreen mode. https://bugzilla.redhat.com/526952 virt-manager locks up frequently using remote connection Problem reported by both Matt Booth and Chris Wright. https://bugzilla.redhat.com/527714 Entering a non-numeric capacity creates a volume with 10000 MB Another minor virt-manager quirk. https://bugzilla.redhat.com/530448 RFE: virt-manager should provide a means to refresh storage pools virsh has a 'refresh-pool' operation, but virt-manager doesn't expose it. https://bugzilla.redhat.com/529956 16x16 icon in Virtual Machine Details window https://bugzilla.redhat.com/529954 virt-manager dialog buttons do not follow the HIG https://bugzilla.redhat.com/529955 virt-manager host details window does not honor button_have_icons setting https://bugzilla.redhat.com/529657 virt-manager accidentally showing tabs in VM details https://bugzilla.redhat.com/530096 virt-manager host details window does not need a menu Another bunch of reports of virt-manager UI warts from Michael Monreal. https://bugzilla.redhat.com/529217 virt-manager storage pool window appears behind dialog This annoyance in Fedora 11 reported by James Laska is fixed upstream, but needs backporting. https://bugzilla.redhat.com/530178 virt-manager right-click menu for connections should have a delete item Yet another UI quirk, this time reported by Gianluca Cecchi. https://bugzilla.redhat.com/529842 virt-manager: Storage browser should change window title when choosing ISO "Choose Volume" was causing some confusion which could be avoided if it was changed to "Choose ISO". == xen == https://bugzilla.redhat.com/523941 kernel 2.6.31-1[24].fc12 doesn't boot in xen PV guest on 64b host Andrew Jones is busy debugging this and has successfully bisected the problem down to a specific upstream commit. Some debate in the bug as to whether we'd actually slip the release for it. In the end, it turned out to be fixed by 2.6.31.5 for unknown reasons. https://bugzilla.redhat.com/531311 2.6.32 boot fails as Xen PV guest with stackprotector Andrew also uncovered this issue with the Fedora 13 kernel. https://bugzilla.redhat.com/526627 F11 Xen DomU unstable (2.6.30) Reporters seeing consistent Fedora 11 Xen guest lockups, seemingly dependant on the clock source. https://bugzilla.redhat.com/521800 kernel backtrace: possible recursive locking detected on Xen domU A strange recursive locking warning under Xen which apparently doesn't make much sense. Resolved Bugs ============= == misc == https://bugzilla.redhat.com/530015 Add libvirt to Virtualization group Somehow when we split libvirt-client from libvirt, we failed to make sure that libvirt was still being pulled in by the Virtualization package group. https://bugzilla.redhat.com/523914 Mouse does not move in PV Xen guest under RHEL-5.4 Wrangling of this complex blocker bug continued. Peter Hutterer (the evdev maintainer) came up with a way to allow evdev to be configured so it works like before. The fixes are now in rawhide. == kernel == https://bugzilla.redhat.com/524229 Local migration of kvm guest fails in Fedora12 Alpha Marcelo thought this was a pvmmu issue fixed in 2.6.31, but apparently not. After some more testing Marcelo now suspects it's a kvmclock related issue fixed by Glauber upstream. This is now fixed in rawhide. https://bugzilla.redhat.com/526869 Add virtio_blk support cache flush (VIRTIO_BLK_F_FLUSH) Justin has pulled this 2.6.32 patch into Fedora 12. https://bugzilla.redhat.com/525890 Backport virtio patches for optimised virtio-net We've decided to wait until 2.6.32 for these patches. == qemu == https://bugzilla.redhat.com/531419 qemu issue with non-virtio NICs receiving heavy traffic volumes Reported by Scott Tsai on the qemu-devel upstream list, patch was pulled into Fedora 12. https://bugzilla.redhat.com/524734 KVM guest ext3 errors at shutdown when using virtio and a qcow2 backing file Kevin debugged this, posted a patch upstream suitable for stable and we pulled it into F-12. https://bugzilla.redhat.com/528901 qemu-kvm : msrs[] array in kvm_arch_save_regs() too small / may cause stack corruption Ulrich Obergfell provided this excellent analysis of a savevm segfault in Fedora 11. Eduardo sent a fix upstream. https://bugzilla.redhat.com/503156 qemu VNC :: xterm inside VM shows garbled text The version of qemu in Fedora 11 appears to have a broken VNC CopyRect encoding implementation and this has been seriously annoying some users. In the absence of a confirmed fix, we've just disabled the encoding for now. https://bugzilla.redhat.com/527653 /etc/init.d/ksm: line 69: [: -eq: unary operator expected Minor bug with the ksm initscript now fixed in F-12. https://bugzilla.redhat.com/527087 qemu-common postinstall scriptlet uses getent before it's installed Bruce Jerrick points out this packaging error. Fixed in F-12 now. https://bugzilla.redhat.com/526381 qcow2 performance bad under i/o load CLOSED NOTABUG since these were compressed qcow2 images. https://bugzilla.redhat.com/530059 VDE support in qemu? CLOSED CANTFIX until someone packages VDE. https://bugzilla.redhat.com/530136 guestfwd option doesn't allow supplementary chardev options like server,nowait Rich Jones points out a problem with the -net user,guestfwd= option, but since libguestfs doesn't need it anymore and the new -chardev option upstream solves the problem, this is now marked CLOSED UPSTREAM. https://bugzilla.redhat.com/528678 qemu-img does not correctly roundtrip data for VPC image format This minor issue was caught by the libvirt-TCK test suite. Kevin reports that this is NOTABUG, but rather a limitation with the vpc file format. https://bugzilla.redhat.com/526776 Add rtl8209 to gpxe-roms-qemu https://bugzilla.redhat.com/526781 Add rtl8209 to etherboot-zroms-kvm https://bugzilla.redhat.com/526243 Guest PXE booting doesn't work when using ne2k_pci NIC model https://bugzilla.redhat.com/526777 Guest PXE booting doesn't work when using ne2k_pci NIC model Fixed this in F-11 and F-12. == libvirt == https://bugzilla.redhat.com/523158 virsh restore causes libvirtd segfault virExec() Dan Berrange identified the recent change that caused this regression. Fixed now in Fedora 12. https://bugzilla.redhat.com/528162 Fix memory leak in libvirtd in F12 An upstream fix which needed backporting to F12. https://bugzilla.redhat.com/526769 libvirt logrotate settings should roll over weekly instead of daily https://bugzilla.redhat.com/531030 log file sized increased by "compression" Two requests to fix logrotate's handling of libvirt's qemu log files. Fixed for F-12. https://bugzilla.redhat.com/517157 libvirt fails to start guest on NFS even when sebool virt_use_nfs is on We never got around to backporting the fix for this to F-11 and other users had been reporting it. Now fixed in a F-11 update. https://bugzilla.redhat.com/524732 org.libvirt.manage policy kit denial This was due to there being no PolicyKit authentication agent running in the session. == virt-manager == https://bugzilla.redhat.com/461962 virt-manager should support logical pool enumeration Creating a lvm volume group using virt-manager is confusing, but Cole has fixed that upstream now. https://bugzilla.redhat.com/440591 virt-manager: Add 'double click to connect' hint to connection row tooltip Another UI annoyance fixed by Cole upstream. https://bugzilla.redhat.com/526111 virt-manager add storage volume allows an empty name field https://bugzilla.redhat.com/526077 virt-manager toolbar buttons should have tooltips https://bugzilla.redhat.com/524083 virt-manager storage "Allocation" field can be set higher than "Max Capacity" Fixed by virt-manager-0.8.0-7.fc12 == xen == https://bugzilla.redhat.com/521806 Unowned directory /usr/lib64/fs https://bugzilla.redhat.com/496135 missing dependency on PyXML Justin Forbes fixed both of these F-12 xen packaging bugs. https://bugzilla.redhat.com/531429 libvirt failed to disconnect and double free crashes on F12/rawhide with Xen Pasi Karkkainen requested a couple of libvirt xen driver fixes from 0.7.2 be backported to Fedora 12.

1 0

ANNOUNCE: Rawhide virt repo for F11 users
by Mark McLoughlin 30 Oct '09

30 Oct '09

Hey, The idea[1] was discussed here before, so I'll keep this short. We've set up a repository for people running Fedora 11 who would like to test the rawhide/F12 virt packages. To use it, do e.g. $> cat > /etc/yum.repos.d/fedora-virt-preview.repo << EOF [rawvirt] name=Virtualization Rawhide for Fedora 11 baseurl=http://markmc.fedorapeople.org/virt-preview/f11/$basearch/ enabled=1 gpgcheck=0 EOF $> yum update At the moment, it contains the F-12 versions of libvirt and qemu, but as F-12 development continues, it will contain more. I'll send periodic mails to the list detailing the latest updates. Also, this is very much a work-in-progress. The TODO list includes: - get createrepo installed on fedorapeople.org - include debuginfo packages in the repo (need more quota) - find a better location than markmc.fedorapeople.org - enable package builders to upload directly to the repo - automatically do preview build/upload for devel/ builds - allow building against preview packages; e.g. for language bindings Comments most welcome. Help with the TODO list is even more welcome :-) Thanks, Mark. [1] - https://www.redhat.com/archives/fedora-virt/2009-April/msg00154.html

9 76

Strange pygrub issue
by Orion Poplawski 30 Oct '09

30 Oct '09

I've got a strange problem with one of my Fedora 11 guests on CentOS 5.4 host. pygrub does not report the current grub menu, but an old (previous) one. So the system boots to an old kernel, despite everything seeming to be up to date in the guest. Any ideas? -- Orion Poplawski Technical Manager 303-415-9701 x222 NWRA/CoRA Division FAX: 303-415-9702 3380 Mitchell Lane orion(a)cora.nwra.com Boulder, CO 80301 http://www.cora.nwra.com

2 1

Trying to move VM efficiently, grub-install, guestfish
by Dennis Jacobfeuerborn 29 Oct '09

29 Oct '09

Hi, What I'm currently experimenting with is how to move a VM between two hosts efficiently. Specifically we are talking Linux guest with virtual disk sizes of 150G of which only about 10-20G are used. Doing a plain "dd" takes a lot of time and is kind of wasteful given that 90% of the logical volume used doesn't contain relevant data. So right now I'm trying to re-create the lvm/filesystem structure from scratch on the target system and then copying the data using tar from the source system. This looks good so far but I cannot start the guest on the target because the boot-sector is missing. I tried using guestfish as described here: http://rwmj.wordpress.com/2009/05/11/rescuing-a-failed-vm-install-with-gues… but when I issue the "grub-install / /dev/sda" all I get is: "libguestfs: error: grub_install: grub-install: /dev/sda does not have any corresponding BIOS drive." I also tried copying the MBR using "dd bs=448 count=1 ..." but that only results in the output of "GRUB" on the target system and then nothing happens. What is the best way to copy/initialize the MBR so the target system boots again? Another method I intend to test is to copy the entire volume with "dd" while the VM is still running then shutting the VM down and using "rsync" to do the final fixup on the logical volumes. That of course assumes that rsync actually works on device nodes which I'm not so sure about. Has anyone experience with this kind of thing? (Just to be clear I'm not talking about cloning the VM requiring making changes to the files in the VM but just simply moving the VM to another host with as little down-time as possible.) Regards, Dennis

2 3

Got Windows guests?
by Richard W.M. Jones 29 Oct '09

29 Oct '09

Hello fellow Fedora, libvirt and libguestfs users, If you have any Windows guests, then you can help Fedora to support Windows guests better by spending a few minutes testing the Windows Registry feature we just added to libguestfs 1.0.75. You will need: - A Windows NT/200x/XP/Vista/7/... guest - Fedora 12 or Fedora Rawhide host - libguestfs-tools >= 1.0.75 (from updates or http://koji.fedoraproject.org/koji/packageinfo?packageID=8391 ) - a few minutes of your time The tests: (1) Run the virt-win-reg commands shown in the following web page, where "MyWinGuest" should be replaced with the name of the Windows guest as known to libvirt: http://libguestfs.org/virt-win-reg.1.html#examples Do the commands run without any errors? Does the output look sensible? If you have several Windows guests, please try as many different sorts as possible! (2) Download the registry binary files and try to convert them to XML: guestfish -i MyWinGuest --ro <<'EOF' download win:\windows\system32\config\software software download win:\windows\system32\config\system system download win:\windows\system32\config\sam sam download win:\windows\system32\config\security security EOF hivexml software > software.xml hivexml system > system.xml hivexml sam > sam.xml hivexml security > security.xml Do those commands run without error? If there's an error, try adding the hivexml -k option. Does the XML look complete? (Try running the XML through tidy -xml -indent -quiet < foo.xml | less ) I hope you don't find any bugs, but if you do: Send a reply to this message, or report a bug in Bugzilla: https://bugzilla.redhat.com/enter_bug.cgi?component=libguestfs&product=Virt… It's useful to include the following details: HIVEX_DEBUG=1 hivexml regfile 2>&1 > log.out The Registry file itself that is failing (but note that Registry files can contain sensitive data). It's also useful to have positive feedback ("it worked!"). Thanks for any testing you can give, and if you have any other suggestions for handling Windows guests from Fedora, please let me know. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://et.redhat.com/~rjones/virt-df/

1 0

F-12 host, can't acquire DHCP lease in guest
by Andrew Overholt 29 Oct '09

29 Oct '09

Hi, I installed rawhide/F-12 x86_64 (network using NetworkManager) and installed and started libvirtd. I then created both F-12 i686 and F-11 i686 guests which went fine. However, neither guest is able to acquire a DHCP lease. Is there some sort of configuration that I didn't do? I don't recall having to do anything special with F-11 as a host. Relevant information below. I can file a bug if that's preferred. Thanks, Andrew ------------------------------------------------------------------------ # virsh net-list --all Name State Autostart ----------------------------------------- default active yes # brctl show bridge name bridge id STP enabled interfaces virbr0 8000.6e8e6e10bdb6 yes vnet0 vnet1 # sysctl net.bridge.bridge-nf-call-iptables net.bridge.bridge-nf-call-iptables = 1 # iptables -L -v -n Chain INPUT (policy ACCEPT 0 packets, 0 bytes) pkts bytes target prot opt in out source destination 2129K 2744M ACCEPT all -- * * 0.0.0.0/0 0.0.0.0/0 state RELATED,ESTABLISHED 3 1070 ACCEPT icmp -- * * 0.0.0.0/0 0.0.0.0/0 7 420 ACCEPT all -- lo * 0.0.0.0/0 0.0.0.0/0 0 0 ACCEPT tcp -- * * 0.0.0.0/0 0.0.0.0/0 state NEW tcp dpt:22 2669 335K REJECT all -- * * 0.0.0.0/0 0.0.0.0/0 reject-with icmp-host-prohibited Chain FORWARD (policy ACCEPT 0 packets, 0 bytes) pkts bytes target prot opt in out source destination 9 2952 ACCEPT all -- * * 0.0.0.0/0 0.0.0.0/0 PHYSDEV match --physdev-is-bridged 0 0 REJECT all -- * * 0.0.0.0/0 0.0.0.0/0 reject-with icmp-host-prohibited Chain OUTPUT (policy ACCEPT 1439K packets, 1128M bytes) pkts bytes target prot opt in out source destination # ps -ef | grep dnsmasq nobody 3148 1 0 09:43 ? 00:00:00 /usr/sbin/dnsmasq --strict-order --bind-interfaces --pid-file=/var/run/libvirt/network/default.pid --conf-file= --listen-address 192.168.122.1 --except-interface lo --dhcp-range 192.168.122.2,192.168.122.254 root 20983 20857 0 13:23 pts/5 00:00:00 grep dnsmasq # ifconfig -a eth0 Link encap:Ethernet HWaddr 00:23:54:9E:57:84 inet addr:10.15.16.87 Bcast:10.15.16.255 Mask:255.255.255.0 inet6 addr: fe80::223:54ff:fe9e:5784/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:2414348 errors:0 dropped:0 overruns:0 frame:0 TX packets:2144264 errors:0 dropped:0 overruns:0 carrier:2 collisions:0 txqueuelen:1000 RX bytes:2888713375 (2.6 GiB) TX bytes:1065980118 (1016.5 MiB) Interrupt:29 lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 inet6 addr: ::1/128 Scope:Host UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:58553 errors:0 dropped:0 overruns:0 frame:0 TX packets:58553 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:256208705 (244.3 MiB) TX bytes:256208705 (244.3 MiB) virbr0 Link encap:Ethernet HWaddr 6E:8E:6E:10:BD:B6 inet addr:192.168.122.1 Bcast:192.168.122.255 Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:85 errors:0 dropped:0 overruns:0 frame:0 TX packets:30 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:19960 (19.4 KiB) TX bytes:6015 (5.8 KiB) vnet0 Link encap:Ethernet HWaddr 6E:8E:6E:10:BD:B6 inet6 addr: fe80::6c8e:6eff:fe10:bdb6/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:14 errors:0 dropped:0 overruns:0 frame:0 TX packets:3410 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:500 RX bytes:3204 (3.1 KiB) TX bytes:180086 (175.8 KiB) vnet1 Link encap:Ethernet HWaddr 92:AE:24:5C:77:7E inet6 addr: fe80::90ae:24ff:fe5c:777e/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:15 errors:0 dropped:0 overruns:0 frame:0 TX packets:3211 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:500 RX bytes:3546 (3.4 KiB) TX bytes:167128 (163.2 KiB) # cat /proc/sys/net/ipv4/ip_forward 1 I've reloaded the libvirtd service, restarted it, rebooted the guests, etc.

2 2

Porting existing virtual machines to newer Fedora
by Gianluca Sforna 27 Oct '09

27 Oct '09

I'm installing Fedora 12 (full install, no upgrade) on my laptop where I have some virtual machines configured via F11's virt-manager. After installation I'd like to recover from a backup the existing virtual machines; is there any documentation on the proper/recommended/safer method for doing the job? TIA G. -- Gianluca Sforna http://morefedora.blogspot.com http://www.linkedin.com/in/gianlucasforna

3 3

Re:[fedora-virt] Request for confirmation/assistance
by Gianluca Cecchi 26 Oct '09

26 Oct '09

On Fri, 23 Oct 2009 11:26:33 -0400 Gene Czarcinski wrote: > The problem appears to be related to "sound" which does not work currently for > a guest and which I do not use. Hello, sorry Gene, can I jump in and ask for having a way to disable sound (aka not having a sound card through the hardware pool components) for a guest during its creation/installation? When you use Qemu/KVM to host several servers, I think that usually you don't want sound at all..... Thanks and let me know if I have to post a bug as enhancement.... if this is not already planned as a feature. Gianluca

1 0

State of KVM paravirtualization
by Giovanni Tirloni 26 Oct '09

26 Oct '09

Hello, I'm trying to understand where KVM stands regarding paravirtualization. I've seen the patches from 2007 by Ingo Molnar (http://people.redhat.com/mingo/kvm-paravirt-patches/) but couldn't find anything newer. Are they integrated to the mainstream kernel? The reason I'm looking at that is because I've a server at home which doesn't have hardware-assisted virtualization support and running qemu in full virtualization mode is a bit slow. I've tried Xen and it sure is fast but I would like to experiment with KVM. Thank you, -- Giovanni.

5 6

Using the virtio drivers in Windows XP setup
by Richard Hughes 23 Oct '09

23 Oct '09

This may be useful to somebody. To use the virtio drivers in windows XP setup, you have to do the following: * create a 1.44Mb image file * mount it by loopback * format it with vfat * copy the Install/Xp/x86/viostor.sys, Install/Xp/x86/wnet.inf, and the attached txtsetup.oem file to the root of the mounted image * umount the loop device * attach the floppy image as a floppy storage element in the VM's details pane * boot the VM * remember to press F6 when booting the windows xp setup and select the VirtoIO device. The txtsetup.oem was just cobbled together after reading various sources online, and hacking values out of the supplied inf file. Richard. p.s. I'm still unable to install XP using ide, scsi or virtio drivers as it gives the message "Setup was unable to format the partition. The disk maybe damaged." -- any ideas welcome.

4 3

Request for confirmation/assistance
by Gene Czarcinski 23 Oct '09

23 Oct '09

I have a qemu-kvm problem which causes a qemu-kvm guest to "suddenly" crash: https://bugzilla.redhat.com/show_bug.cgi?id=528754 This has happened to me at least 5 times on a F12-rawhide host running qemu-kvm-0.11.0-6.fc12.x86_64 The guests (which crashed) have been F11 and F12 with (at the time) all updates applied. The problem appears to be related to "sound" which does not work currently for a guest and which I do not use. I do not know what action on my part "causes" the problem. When the problem occurs, there have been (IIRC) two guests running (F11 and F12). When the problem happens, the guest console suddenly becomes a white screen with "Guest not running". There are messages in /var/log/messages which say that qemu-kvm has had a segfault. OK, has anyone else seen anything like this? So far, there has been insufficient information to give the developers, etc. any idea of where the problem is. I suspect it would be helpful if someone else had the problem. Gene

1 0

virbr0 messing with iptables rules?
by Kenni Lund 23 Oct '09

23 Oct '09

Hi I just did a full system update on my F11 server, but after a reboot, new rules were appended to my iptables setup. The iptables and ip6tables services are both disabled: # chkconfig |grep ip.*tables ip6tables 0:off 1:off 2:off 3:off 4:off 5:off 6:off iptables 0:off 1:off 2:off 3:off 4:off 5:off 6:off I set my iptables rules in a custom firewall script in /etc/rc.local, which starts out by flushing all rules. Eg. if I run the script manually after boot, it will "fix" the issue. The extra rules appended are: ----- Chain INPUT (policy DROP) target prot opt source destination ACCEPT udp -- anywhere anywhere udp dpt:domain ACCEPT tcp -- anywhere anywhere tcp dpt:domain ACCEPT udp -- anywhere anywhere udp dpt:bootps ACCEPT tcp -- anywhere anywhere tcp dpt:bootps Chain FORWARD (policy DROP) target prot opt source destination ACCEPT all -- anywhere 192.168.122.0/24 state RELATED,ESTABLISHED ACCEPT all -- 192.168.122.0/24 anywhere ACCEPT all -- anywhere anywhere REJECT all -- anywhere anywhere reject-with icmp-port-unreachable REJECT all -- anywhere anywhere reject-with icmp-port-unreachable ----- I suppose that these rules are related to libvirt, since I have a virbr0 interface with the IP-address 192.168.122.1. Apparently the update changed something, so these firewall rules are appended after /etc/rc.local runs my custom firewall script. What is the correct solution to this? In general, isn't it a bad design decision to have a service mess with the iptables rules, instead of doing this through the iptables/ip6tables services? I would not expect other services to mess with my rules, when I explicitly disabled the build-in iptables services. I only use bridged networking for my virtual machines, so I don't suppose that I need the virbr0 interface after all. Can I disable it somewhere? Best Regards Kenni Lund

2 1

Re: [fedora-virt] f12-beta geust install fails on f11
by Dale Bewley 22 Oct '09

22 Oct '09

----- "Cole Robinson" <crobinso(a)redhat.com> wrote: > On 10/22/2009 01:13 AM, Dale Bewley wrote: > > Host is f11 with virt-manager-0.7.0-7.fc11.x86_64 > > > > Using the guest creation wizard, I tried to install f12-beta. I > pointed at a http mirror and allowed virt-manager to detect the OS and > variant, but it failed with: > > > > Unable to complete install: 'OS variant 'fedora12-beta; does not > exist in our dictionary for OS type 'linux'' > Can you try pulling virtinst from updates testing?: > > yum --enablerepo=updates-testing update python-virtinst > > There was a relevant fix that I believe hasn't been pushed to stable > yet. Please report the results here. > Yes that seems to work. It detects f12-beta as OS Generic. I'll go give it some Bodhi karma. Thanks, Cole.

2 1

f12-beta geust install fails on f11
by Dale Bewley 22 Oct '09

22 Oct '09

Host is f11 with virt-manager-0.7.0-7.fc11.x86_64 Using the guest creation wizard, I tried to install f12-beta. I pointed at a http mirror and allowed virt-manager to detect the OS and variant, but it failed with: Unable to complete install: 'OS variant 'fedora12-beta; does not exist in our dictionary for OS type 'linux'' I went back and selected F11, and it works fine so far of course. Would it make sense if the auto-detect could sense alpha and beta releases then punt to a default such as the most recent release of the distribution or some other common denominator? -- Dale Bewley - Unix Administrator - Shields Library - UC Davis GPG: 0xB098A0F3 0D5A 9AEB 43F4 F84C 7EFD 1753 064D 2583 B098 A0F3

2 1

virt-manager and ssh with non root user
by Gianluca Cecchi 21 Oct '09

21 Oct '09

Hello, using F11 + rawvirt repo. I would like to connect from my virt-manager from this F11 to a RH EL 5.4 xen Hypervisor. Is it possible? On the RH EL host, root access via ssh is prevented and I don't want to use a CA certficate at the moment. Also I cannot export DISPLAY env so I cannot directly ssh to the host and run its virt-manager program.... I tried to edit /etc/libvirtd/libvirtd.conf on rh el 5.4 host adding unix_sock_group = "mygroup" and then trying to connect with one user of this group. Is this setting correct as I intended it? When in virt-manager of F11 I try to create new connection, it seems I'm only able to create a session of type xen+ssh://root@myrhel54host/ and not of type xen+ssh://myuser@myrhel54host Any hints? Also, as in virt-manager this connection fails to create (root prevented from ssh directly as said above), I would like to delete it... but I didn't find an option for this; In virt-manager only "connect" and "details" options are available.... these are messages in virsh session when setting LIBVIRT_DEBUG=1 virsh # connect xen+ssh://unix@myrhel54host 17:23:44.911: debug : virConnectClose:1291 : conn=0x65de70 17:23:44.911: debug : virUnrefConnect:257 : unref connection 0x65de70 1 17:23:44.911: debug : remoteIO:7421 : Do proc=2 serial=3 length=28 wait=(nil) 17:23:44.911: debug : remoteIO:7483 : We have the buck 2 0x67db20 0x67db20 17:23:44.912: debug : remoteIODecodeMessageLength:7032 : Got length, now need 56 total (52 more) 17:23:44.912: debug : remoteIOEventLoop:7347 : Giving up the buck 2 0x67db20 (nil) 17:23:44.912: debug : remoteIO:7514 : All done with our call 2 (nil) 0x67db20 17:23:44.912: debug : virReleaseConnect:214 : release connection 0x65de70 17:23:44.912: debug : virConnectOpenAuth:1273 : name=xen+ssh://unix@myrhel54host, auth=0x7fb212bc40a0, flags=0 17:23:44.912: debug : do_open:1042 : name "xen+ssh://unix@myrhel54host" to URI components: scheme xen+ssh opaque (null) authority (null) server myrhel54host user unix port 0 path (null) 17:23:44.912: debug : do_open:1052 : trying driver 0 (Test) ... 17:23:44.912: debug : do_open:1058 : driver 0 Test returned DECLINED 17:23:44.912: debug : do_open:1052 : trying driver 1 (Xen) ... 17:23:44.912: debug : do_open:1058 : driver 1 Xen returned DECLINED 17:23:44.912: debug : do_open:1052 : trying driver 2 (OPENVZ) ... 17:23:44.912: debug : do_open:1058 : driver 2 OPENVZ returned DECLINED 17:23:44.912: debug : do_open:1052 : trying driver 3 (PHYP) ... 17:23:44.912: debug : do_open:1058 : driver 3 PHYP returned DECLINED 17:23:44.912: debug : do_open:1052 : trying driver 4 (VBOX) ... 17:23:44.912: debug : do_open:1058 : driver 4 VBOX returned DECLINED 17:23:44.912: debug : do_open:1052 : trying driver 5 (ESX) ... 17:23:44.912: debug : do_open:1058 : driver 5 ESX returned DECLINED 17:23:44.912: debug : do_open:1052 : trying driver 6 (remote) ... 17:23:44.912: debug : doRemoteOpen:535 : proceeding with name = xen:// 17:23:44.912: debug : virExecWithHook:609 : ssh -l unix myrhel54host nc -U /var/run/libvirt/libvirt-sock 17:23:44.917: debug : remoteIO:7421 : Do proc=66 serial=0 length=28 wait=(nil) 17:23:44.918: debug : remoteIO:7483 : We have the buck 66 0x6bde80 0x6bde80 17:23:44.924: debug : remoteIOEventLoop:7368 : Giving up the buck due to I/O error 66 0x6bde80 (nil) 17:23:44.924: debug : do_open:1058 : driver 6 remote returned ERROR 17:23:44.924: debug : virUnrefConnect:257 : unref connection 0x65e260 1 17:23:44.924: debug : virReleaseConnect:214 : release connection 0x65e260 error: Failed to connect to the hypervisor error: cannot recv data: Connection reset by peer Thanks, Gianluca

3 3

Re: [fedora-virt] virtio block driver performance versus ide
by Richard W.M. Jones 21 Oct '09

21 Oct '09

On Wed, Oct 21, 2009 at 08:58:10AM -0600, Thomas S Hatch wrote: > Wow, the rotational change greatly speeds up my vms, as much as 15 fold. > > At the risk of sounding like a complete n00b, is there a way to persistently > change the settings in /sys? I am of course familiar with the sysctl.conf > for /proc/sys. Or is this something that should be changed at the driver > level? It has apparently already been changed (see the new kernels in that Bugzilla entry). In any case, you can't make the change permanent, but you can add the commands to /etc/rc.local so they get executed at each guest boot which has just about the same effect. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones libguestfs lets you edit virtual machines. Supports shell scripting, bindings from many languages. http://et.redhat.com/~rjones/libguestfs/ See what it can do: http://et.redhat.com/~rjones/libguestfs/recipes.html

1 0

virtio block driver performance versus ide
by Dale Bewley 21 Oct '09

21 Oct '09

I assumed the virtio block driver to be faster for a kvm guest than ide, but I'm seeing quite the opposite. Is this expected? I normally create guests with a wrapper script around virt-install which creates guests configured to use IDE disks: f11-db1.xml- <emulator>/usr/bin/qemu-kvm</emulator> f11-db1.xml: <disk type='file' device='disk'> f11-db1.xml- <source file='/var/lib/libvirt/images//f11-db1.img'/> f11-db1.xml- <target dev='hda' bus='ide'/> f11-db1.xml: </disk> I am currently creating a guest with the virt-manager wizard which apparently uses virtio by default: f11-archivist.xml- <emulator>/usr/bin/qemu-kvm</emulator> f11-archivist.xml: <disk type='file' device='disk'> f11-archivist.xml- <source file='/var/lib/libvirt/images//f11-archivist.img'/> f11-archivist.xml- <target dev='vda' bus='virtio'/> f11-archivist.xml: </disk> The filesystem formatting and installation are going very very slowly. Host is 2.6.29.6-213.fc11.x86_64 4x quad core. Guests are also fc11 x86_64. BTW, I've updated to virt-manager-0.7.0-7, libvirt-0.6.2-18, qemu-kvm-0.10.6-6, and kernel-2.6.30.8-64 just before creating this guest but have not restarted libvirtd or rebooted the host. -- Dale Bewley - Unix Administrator - Shields Library - UC Davis GPG: 0xB098A0F3 0D5A 9AEB 43F4 F84C 7EFD 1753 064D 2583 B098 A0F3

5 5

f12 guest memory problems?
by Gianluca Cecchi 21 Oct '09

21 Oct '09

Hello, F11 x86_64 + rawvirt with latest updates I have a f12 near beta x86_64 kvm guest where now I'm doing "yum update" in it and basically I'm only in gui session without doing anything else Some strange things: - in virt-manager set 400Mb for VM (both the allocated, change, max allocation values... btw how to use them???) my guest sees only about 330Mb [gcecchi@localhost ~]$ free total used free shared buffers cached Mem: 344212 252944 91268 0 4268 26908 -/+ buffers/cache: 221768 122444 Swap: 950264 179092 771172 With dmesg on guest I see something like this: Memory: 329452k/409536k ...... 79696k reserved The 79696k reserved seems to fill the gap... is this a feature in f12 or based on it being a guest? - As you see, the guest is swapping about 180Mb, and actually it is very slowly responsive. Shouldn't it use some of buffers/cache before swapping so much? Is this expected during a yum update (only 34 packages for an overall of 72MB)? Or problem with F12 itself or virtualization layer? Thanks Gianluca

1 0

Re: Oops
by Mark McLoughlin 21 Oct '09

21 Oct '09

Hi Gene, On Tue, 2009-10-20 at 19:39 -0400, Gene Czarcinski wrote: > F12 beta was available so I downloaded the DVD iso ... that was a waste of > bandwidth since it was identical to F12-Beta RC2. > > Anyway, I had already installed a couple of copies of RC2 into qemu-kvm guests > so it was time to install this on bare-metal. > > My test system already had F12-rawhide installed with all current updates but, > afterall, this is a test of the install ... so I did another install ... my > test system supports four, separate bootable systems (combinations of /boot > partitions and LVM logical volumes for root). > > OK, boot it up and start configuring it, updating (I did a DVD-only install), > etc. Well, one of the things this system does is run qemu-kvm so I had > selected Virtualization during the install. My guest disk images are all on a > separate logical volume which I mount on /.var/lib/libvirt/images but the > system definition/configuration files are separate. When I went to copy them > from the old system I found the was no place to copy them to ... /etc/libvirt/ > was missing. Further checking showed that the libvirt rpm was not installed > ... this is easy enought to fix but it should not have happened ... it should > have been installed when I selected virtualization. Oops, indeed. I've filed this here: https://bugzilla.redhat.com/530015 Thanks for finding this, it's a very serious issue that could have crept by unnoticed. Cheers, Mark.

1 0

#529711 - qemu: could not open serial device 'pty'
by Gianluca Cecchi 20 Oct '09

20 Oct '09

On Mon, 19 Oct 2009 20:15:51 +0200 Gianluca Varisco wrote: > Hi guys, > > Am I the only one having this issue with qemu/KVM and rawhide? > > https://bugzilla.redhat.com/show_bug.cgi?id=529711 <https://bugzilla.redhat.com/show_bug.cgi?id=529711> > > (It is not a SELinux-related problem) > > Cheers, I would advise to also follow what written at link: https://fedoraproject.org/wiki/Reporting_virtualization_bugs In particular contents of virt-manager.log and posting output when setting LIBVIRT_DEBUG=1 btw I presume that also the version of virt-manager is the latest one, correct? HIH, Gianluca

1 0

#529711 - qemu: could not open serial device 'pty'
by Gianluca Varisco 19 Oct '09

19 Oct '09

Hi guys, Am I the only one having this issue with qemu/KVM and rawhide? https://bugzilla.redhat.com/show_bug.cgi?id=529711 (It is not a SELinux-related problem) Cheers, -- Gianluca Varisco, RHCE | Office: +39 02 9737 4652 Red Hat Italia | Fax: +39 02 669 3111 Via Antonio Da Recanate 1 | Mobile: +39 333 574 0934 20124 Milano | eMail: gvarisco(a)redhat.com

1 0

'eject' on a KVM guest
by Matthew Booth 19 Oct '09

19 Oct '09

What does the eject (to eject a CD) command do to a KVM guest? It succeeds, and the result is that the device can no longer be mounted. However, looking at the domain XML reveals it hasn't disconnected the backing file. Matt -- Matthew Booth, RHCA, RHCSS Red Hat Engineering, Virtualisation Team M: +44 (0)7977 267231 GPG ID: D33C3490 GPG FPR: 3733 612D 2D05 5458 8A8A 1600 3441 EA19 D33C 3490

3 4

how to edit dhcp configuration
by Pavel Lisy 19 Oct '09

19 Oct '09

Hello I want to set fixed IP for host (from MAC address) in dhcp. How can I do it through libvirt configuration? There is not possibility set it in virt-manager. I can edit manually ldapnet.xml (ldapnet is name of my network) but: 1. question: Where is right place to edit it: /etc/libvirt/qemu/networks/ldapnet.xml or /var/lib/libvirt/network/ldapnet.xml (after reboot one file replaces other but which direction but I didn't find which direction) 2. question: How can I propagate this changes? Neither /etc/init.d/libvirtd reload /etc/init.d/libvirtd restart changes dnsmasq settings. It is done only during reboot, why? 3. I've found this workaround: * kill dnsmasq process for this settings * start dnsmasq with new settings, ex. .... --dhcp-host 52:54:00:22:a7:76,ldap-m1.ldapnet.tmapy.cz,192.168.241.101 Is it way how to do it through libvirt utility? Pavel -- Pavel Lisy <pavel.lisy(a)gmail.com>

2 1

Re: nagios-virt: configure: error: Cannot find `nagios' program
by Richard W.M. Jones 16 Oct '09

16 Oct '09

On Fri, Oct 16, 2009 at 11:39:38AM +0200, Stephane Grand wrote: > Hello, > > I've found your web page about nagios-virt (http://et.redhat.com/~rjones/nagios-virt/) This tools interesting me to supervise servers and vm. > > Unfortunately, I can't test it. Here is error message I got when I execute "./configure" > > [root] # ./configure > ... > checking for stdint.h... yes > checking for unistd.h... yes > checking libvirt/libvirt.h usability... yes > checking libvirt/libvirt.h presence... yes > checking for libvirt/libvirt.h... yes > checking for virConnectGetCapabilities... yes > checking for nagios... no > configure: error: Cannot find `nagios' program > > I've retried this command with others options but I got the same error message Does the 'nagios' command exist? Is it on the $PATH? Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones libguestfs lets you edit virtual machines. Supports shell scripting, bindings from many languages. http://et.redhat.com/~rjones/libguestfs/ See what it can do: http://et.redhat.com/~rjones/libguestfs/recipes.html

2 1

Re: Re: nagios-virt: configure: error: Cannot find `nagios' program
by Stephane Grand 16 Oct '09

16 Oct '09

Thank you for the answer. Here is what I have done to follow the installation process. [root] # PATH=$PATH:/var/supervision/nagios/bin/ [root] # ./configure --with-nagios-configdir=/var/supervision/nagios/etc [root] # make [root] # ./nagios-virt config PACKAGE_NAME nagios-virt PACKAGE_VERSION 0.2.0 NAGIOS_CONFIGDIR /var/supervision/nagios/etc/ NV_DATADIR /usr/local/share/nagios-virt-0.2.0 ==>But I've got another problem. [root] # ./nagios-virt install Installing virt-library.cfg in /var/supervision/nagios/etc/ ... Installing /var/supervision/nagios/etc//commands.cfg ... Warning: could not install /var/supervision/nagios/etc//commands.cfg This file from the standard Nagios distribution is required in order for nagios-virt to work correctly. Installing virt-hosts.cfg in /var/supervision/nagios/etc/ ... virConnectNumOfDomains: failed Why this command can not install "commands.cfg" file ? Thank you. Stephane ======================================== Date du message : oct. 16 2009, 12:05 PM De : "Richard W.M. Jones" <rjones(a)redhat.com> A : "Stephane Grand" <stephane.grand(a)atosorigin.com>, fedora-virt(a)redhat.com Copie : Sujet : Re: nagios-virt: configure: error: Cannot find `nagios' program On Fri, Oct 16, 2009 at 11:39:38AM +0200, Stephane Grand wrote: > Hello, > > I've found your web page about nagios-virt (http://et.redhat.com/~rjones/nagios-virt/) This tools interesting me to supervise servers and vm. > > Unfortunately, I can't test it. Here is error message I got when I execute "./configure" > > [root] # ./configure > ... > checking for stdint.h... yes > checking for unistd.h... yes > checking libvirt/libvirt.h usability... yes > checking libvirt/libvirt.h presence... yes > checking for libvirt/libvirt.h... yes > checking for virConnectGetCapabilities... yes > checking for nagios... no > configure: error: Cannot find `nagios' program > > I've retried this command with others options but I got the same error message Does the 'nagios' command exist? Is it on the $PATH? Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones libguestfs lets you edit virtual machines. Supports shell scripting, bindings from many languages. http://et.redhat.com/~rjones/libguestfs/ See what it can do: http://et.redhat.com/~rjones/libguestfs/recipes.html ----------------------- Atos Origin Stéphane Grand Open Source Center dcen-net-grand.consultant(a)dgfip.finances.gouv.fr

1 0

USB passthrough and disconnect
by Gianluca Cecchi 15 Oct '09

15 Oct '09

Hello, following the instructions at http://www.redhat.com/archives/fedora-virt/2009-June/msg00182.html I'm able to attach an USB key to my Windows XP SP3 guest. More better, I'm able to attach and use it with the guest already started. Very good! (Actually I don't have SElinux enabled so I skipped what related to it ) The doubt is, when in Windows I then safely remove the key and get the ok msg from the os, if I try to hot remove the USB device in details windows of the guest I receive the message that will be effective after reboot. Is this expected and the hot unplug a feature next to come, or do I miss anything in correct way to do things? My sw versions for host are F11 x86_64 with up2date rawvirt repo: libvirt-0.7.1-11.fc11.x86_64 qemu-kvm-0.11.0-6.fc11.x86_64 virt-manager-0.8.0-7.fc11.noarch Suppose the message is correct behaviour based on current state of things, what would be correct approach if during a vm session I use several usb keys for example? 1) insert first usb key in host 2) add hw of first usb key in details tab of guest 3) use it in guest and at the end disconnect it from inside guest OS 4) remove hw in details tab of guest; confirming that this will overwrite any othe rchanges occured in guest config.... 5) remove key form host usb plug 6) insert second usb key in host 7) add hw of second usb key in details tab of guest 8) use it in guest and at the end disconnect it from inside guest OS 9) ? If I do same as 4 do I loose the action made in 4) ?? Do I have to remove more hw components together only at the end? Any removal will let forget the previous ones? It seems not clear to me this message about " This device could not be removed from the running machine. Would you like to remove the device after the next VM shutdown? Warning: this will overwrite any other changes that require a VM reboot. " This is true for more than one usb key, but also for more than one device in general.... Thanks in advance, Gianluca

2 1

status of guest not well visible when selected
by Gianluca Cecchi 13 Oct '09

13 Oct '09

So all is working like a charm and I have to scale to aesthetics.... ;-) With latest rawvirt repo on F11, when I have a guest selected in virt-manager gui, the current status of it (Running vs Shutoff) is pretty difficult to notice.... Ok, there is the play symbol inside the terminal icon but probably a black color for the state of the guest, at least when selected, would be more readable. Planning to scratch these two hosts of mine from F11 + rawvirt to F12 Beta... any insights/suggestions? Preupgrade option viable in your opinion? Bye and thanks, Gianluca

2 1

Re: [fedora-virt] how to attach to qemu monitor
by Gianluca Cecchi 13 Oct '09

13 Oct '09

On Tue, Oct 13, 2009 at 3:52 PM, Don Dutile <ddutile(a)redhat.com> wrote: > Gianluca Cecchi wrote: > >> I don't remember if it is possible to connect to qemu monitor with >> virt-manager/libvirtd managed guests. >> If so, how can I accomplish this? >> >> Thanks in advance, >> Gianluca >> > do you mean?: > > virsh console <guest-name> > No, not this. Putting correct lines in guest kernel command line I'm able to see what you wrote, directly from virt-manager through View --> Consoles --> Serial 0 I mean qemu monitor itself. The one that you activate inside qemu guest when you run qemu or qemu-kvm and then press Ctrl+ Alt + 2 and you get something like QEMU 0.11.0 monitor - type 'help' for more information (qemu) I understand that it would be risky to make things from the qemu monitor itself bypassing libvirt, and so I asked, just to know Thanks, Gianluca

1 0

how to attach to qemu monitor
by Gianluca Cecchi 13 Oct '09

13 Oct '09

I don't remember if it is possible to connect to qemu monitor with virt-manager/libvirtd managed guests. If so, how can I accomplish this? Thanks in advance, Gianluca

1 0

gPXE and localboot
by Ewan Mac Mahon 13 Oct '09

13 Oct '09

Hi, According to https://fedoraproject.org/wiki/Features/VirtgPXE the switch to gPXE should make PXE 'localboot' (which etherboot didn't support) start working. I've tried setting up a network booting VM on a host running F11 + virt-preview and while it uses gPXE, loading a pxelinux configuration that tries to do localboot doesn't work - gPXE loops and does DHCP again[1]. The same config does work on real hardware, so am I missing a detail somewhere, do I need a full F12 install, or is this not supposed to be working yet? Ewan [1] screenshot here: http://www-pnp.physics.ox.ac.uk/~macmahon/gPXElocalboot.png

2 1

Re: Question about segmentation fault under febootstrap
by Richard W.M. Jones 12 Oct '09

12 Oct '09

On Sat, Oct 10, 2009 at 04:56:35PM +0300, Vitaliy Yermolenko wrote: > I'm a big fun of Red Hat/Fedora distributives, and, have tried your > febootstrap Fedora bootstrapping system, which seems very interesting for me > in order to make own lightweighted Fedora-based Linux distributive. > I'm currently using Fedora 11 with fresh 2.6.31.1 kernel on AMD Phenom II X4 > 945 CPU. But, unfortunately, I have got many segmentation faults during > running ./minimal-filesystem.sh - see below. You should ask questions on one of the proper mailing lists. I've CC'd this to fedora-virt. Segfaults are almost certainly because you're trying to install a guest >= F12 on an F11 host. There's some ABI incompatibility in glibc which causes fakeroot to segfault in this case. Install an F11 guest if you have an F11 host, or upgrade to F12 host. Rich. -- Richard Jones, Emerging Technologies, Red Hat http://et.redhat.com/~rjones libguestfs lets you edit virtual machines. Supports shell scripting, bindings from many languages. http://et.redhat.com/~rjones/libguestfs/ See what it can do: http://et.redhat.com/~rjones/libguestfs/recipes.html

1 0

libvirtd and public access to guests
by Pavel Lisy 09 Oct '09

09 Oct '09

Hello I've started playing with libvirt and I have question? What is proper way to make guest accessible from net. I have mode=nat /var/lib/libvirt/network/default.xml. libvirtd makes this rules in FORWARD chain -A FORWARD -d 192.168.231.0/24 -o virbr0 -m state --state RELATED,ESTABLISHED -j ACCEPT -A FORWARD -s 192.168.231.0/24 -i virbr0 -j ACCEPT -A FORWARD -i virbr0 -o virbr0 -j ACCEPT -A FORWARD -o virbr0 -j REJECT --reject-with icmp-port-unreachable -A FORWARD -i virbr0 -j REJECT --reject-with icmp-port-unreachable -A FORWARD -j REJECT --reject-with icmp-host-prohibited If I add iptables -I FORWARD -i eth0 -o virbr0 -j ACCEPT guests are accessible My question is: Is is possible write this somewhere to configuration? I've tried to put it in /etc/sysconfig/iptables but it libvirtd put his rules before mine. I've found two directories /var/lib/libvirt/iptables/filter /var/lib/libvirt/iptables/nat I suppose I can write my rules here but I haven't find any docs about format. Can somebody help me with it? Pavel

3 3

user qemu has no access to usb devices
by Tom Horsley 09 Oct '09

09 Oct '09

I don't know if something is busted in my rawhide setup, or if this is just a real problem with the new scheme of running as user qemu, but user qemu (on my system at least), has no access to physical usb devices. I had to change the qemu.conf file to tell it to run as root before it could attach a usb device to my windows XP virtual machine. See https://bugzilla.redhat.com/show_bug.cgi?id=524723#c5

2 1

<Hostdevice> parameters
by Andrés García 08 Oct '09

08 Oct '09

Hi, I want to set up wo virtual WinXP machines to act as licence managers for a couple of CAD/CAM programs. With the first one, everything went well, but when I got the second one to work, the virtual machines got confused, as if they didn't know which was their dongle. I think it is because in their definition files there is something like: ... <hostdev mode='subsystem' type='usb'> <source> <vendor id='0x1234'/> <product id='0xbeef'/> </source> </hostdev> ... The vendor and product ids are the same in both cases. I saw in the docs that I can't change it to: <hostdev mode='subsystem' type='sub'> <source> <address bus='0x06' slot='0x02' function='0x0'/> </source> </hostdev> 'bus' and 'slot' I can get with 'lsusb', but what does 'function' mean? Thanks, Andrés

2 2

Should python-virtinst %require libvirt?
by James Laska 08 Oct '09

08 Oct '09

Greetings, Quick question about virt package dependencies. The Rawhide Acceptance Install Test (which does a virt install) [1] first installs the required packages by doing: # yum -y install qemu-kvm python-virtinst pax It seems that installing python-virtinst used to also pull libvirt into the transaction. However, this is no longer the case [2]. Should the deps chain when installing python-virtinst also install libvirt? Or is it expected that if you wish to run your own virt daemon, you should install libvirtd? Thanks, James [1] http://git.fedorahosted.org/git/?p=autoqa.git;a=blob;f=tests/rats_install/r… [2] rats_install test output - http://pastie.org/647021

2 2

question on virt-install (and partly anaconda)
by Kashyap Chamarthy 08 Oct '09

08 Oct '09

Hi, I run something like this, for un-attended automated guest installs. =============================================================================================== #!/bin/bash domname=yellow location=http://10.65.x.x/fedora/linux/releases/11/Fedora/x86_64/os/ vmimage="/var/lib/libvirt/images/$domname.img" echo "Creating domain $domname" echo "Image is here $vmimage" echo "Location of the OS sources $location" virt-install --connect=qemu:///system \ --network=bridge:br0 \ --extra-args="ks=http://10.65.y.y/fedora-minimal.ks console=tty0 console=ttyS0,9600" \ --name $domname \ --file=$vmimage \ --file-size=8 \ --ram 1024 \ --accelerate \ --nonsparse \ --location=$location \ ============================================================================================ now here, -- once the install is finished and guest reboots, a console does /not/ show up. Because, as the "console=tty0 console=ttyS0,9600" args from the script are not picked by the anaconda of guest while booting. -- A console shows up only when I manually edit the guest machine's kernel args and append "console=tty0 console=ttyS0,9600" , /then/ boot the guest - now console shows up happily. My question is : shouldn't the anaconda of the guest machine pick up automatically "console=tty0 console=ttyS0,9600" in the --extra-args mentioned in the virt-install above? Is this the expected behaviour? or shall I file a bug. Regards, Kashyap

2 3

Providing access to a single VM
by Dennis Jacobfeuerborn 07 Oct '09

07 Oct '09

Hi, is there a way to provide virt-viewer access to a single VM? After some googling I can only find ways to provide access to a whole virt-manager instance but I'd like to create VMs for users an only give them access to their own virtual machines? Is this possible? Regards, Dennis

2 1

Domain doesn't show up
by Tim Richardson 07 Oct '09

07 Oct '09

I'm trying to install xp virtual machine from an .iso on the disk, I have a virtual machine called localhost with qemu as ID. It has a storage area at /var/lib/libvirt/images, i'm right clicking on that and choosing new then following the wizard but it keeps failing whilst creating the domain. Is anyone familiar with this scenario or have any suggestions? regards Tim Unable to complete install '<class 'libvirt.libvirtError'> internal error Domain microXP didn't show up Traceback (most recent call last): File "/usr/share/virt-manager/virtManager/create.py", line 1501, in do_install dom = guest.start_install(False, meter = meter) File "/usr/lib/python2.6/site-packages/virtinst/Guest.py", line 541, in start_install return self._do_install(consolecb, meter, removeOld, wait) File "/usr/lib/python2.6/site-packages/virtinst/Guest.py", line 633, in _do_install self.domain = self.conn.createLinux(install_xml, 0) File "/usr/lib64/python2.6/site-packages/libvirt.py", line 974, in createLinux if ret is None:raise libvirtError('virDomainCreateLinux() failed', conn=self) libvirtError: internal error Domain microXP didn't show up '

2 1

suggestion: a service for tuning ksm according to memory load
by Dan Kenigsberg 06 Oct '09

06 Oct '09

If you are running a number of qemu-kvm's with similar guests, you can gain a lot of memory by using ksm properly. An unattended host running a variable number of qemu-kvm's needs to tune ksm automatically, since when memory is tight, it's better to spend more cpu on merging pages. In more relaxed cases, it's just a waste of time. The attached service tries to do just that. It monitors how much memory is used by qemu-kvm processes, and starts ksm when a threshold is passed. Ksm usually manages to free up some memory. As long as memory used by qemu is above the defined threshold, ksm tries harder and harder to share memory pages (up to a limit). This may happen if a guest starts working and consumes new memory. If there's enough free memory, ksm cools down. Ksmd service has the usual start/status/stop verbs, and an additional one: signal. One should use that verb just after one starts a new qemu-kvm process or just after such process dies, to let ksm adjust immediately. Comments and suggestion are welcome. Thanks, Dan.

4 13

virt-viewer windows port
by anuj rampal 06 Oct '09

06 Oct '09

Hi all, I have build virt-viewer-0.2.0 for windows. now when I tried to run that "virt-viewer.exe" for the 1st time. it gave me this error: (virt-viewer.exe:8464): Pango-WARNING **: `Z:\usr\i686-pc-mingw32\sys-root\mingw \lib\pango\1.6.0\modules\pango-basic-win32.dll': The specified module could not be found. To solve this I had to rename one of my Disk to Z and the copied the pango-basic-win32.dll at the specified path. now it didnt give me any error. But what happens now is: 1. virt-viewer.exe -c qemu+tcp://192.168.1.106/session 1 ; where 192.168.1.106 is the ip address of my machine where qemu is running and 1 is the id of the VM. 2. A blank screen opens up with "virt-viewer.exe" written on its top and then that program does not responds. Could someone please help me out with this. Thanks in advance Regards Anuj

3 16

[ANNOUNCE] virt-top 1.0.4
by Richard W.M. Jones 05 Oct '09

05 Oct '09

virt-top is the sane replacement for xentop. Virt-top looks and acts like the familiar top(1) command, displays virtual machines, and uses libvirt so it works with just about every virtualization system out there. It also has cool features for sysadmins, like you can use it to log stats into a database or spreadsheet. This version fixes a few long-standing bugs, particularly in the handling of languages other than English. We've also tidied up the source and moved to using git for source control. Home page: http://et.redhat.com/~rjones/virt-top/ Source: http://et.redhat.com/~rjones/virt-top/files/ Repo: http://git.et.redhat.com/?p=virt-top.git;a=summary My blog: http://rwmj.wordpress.com/tag/virt-top/ There's already a build in Fedora 13 which will probably work in earlier versions of Fedora: http://koji.fedoraproject.org/koji/packageinfo?packageID=6211 Rich. -- Richard Jones, Emerging Technologies, Red Hat http://et.redhat.com/~rjones virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://et.redhat.com/~rjones/virt-df/

1 0

virt-manager and manually created bridges
by Sven Lankes 03 Oct '09

03 Oct '09

My home-server is running rawhide. To get some virtual machines working on it I did: * disable Networkmanager * setup a br0-Device that includes eth0 * disable the libvirt default network virbr0 using virsh net-autostart default --disable When I now try to install a new vm using virt-manager this fails: I'm asked (under advanced options) to choose a network - the only network available ist the (deactivated) virbr0 network. When I then try to proceed anyway I am told that the virbr0 is inactive and would I like to start it. Choosing no doesn't allow me to finish the installation wizard. When I edit an xml definition manually to add the br0-Bridge as network device things work fine. Shouldn't this be possible from within virt-manager? -- sven === jabber/xmpp: sven(a)lankes.net

2 1

Fedora virt status
by Mark McLoughlin 02 Oct '09

02 Oct '09

Fedora 12 ========= The Fedora 12 release is drawing closer and closer. Beta is on its way out the door and release candidate composes will begin in less than four weeks time. Helping Out =========== Here are three ways you could help out with getting F12 into great shape: 1) Test, test, test and file bugs. The more stuff you break, the more will get fixed. Maybe you could even write some test cases; for examples, just take a look at: https://fedoraproject.org/wiki/Test_Day:2009-09-17_Virtualization https://fedoraproject.org/wiki/Test_Day:2009-05-07_Virtualization 2) Follow the action in bugzilla. See this wiki page for how to sign up to receive all bugzilla emails relating to Fedora virtualization: https://fedoraproject.org/wiki/Virtualization_bugs 3) Take a look at some existing bugs and see if you can help out getting them closed. A good place to start is the F12 target list: https://bugzilla.redhat.com/showdependencytree.cgi?id=F12VirtTarget&hide_re… But be warned, you may quickly find yourself becoming a package co-maintainer! :-) Test Day ======== Earlier this month, a dedicated bunch of folks got together on irc for the day to see what they could break: https://fedoraproject.org/wiki/Test_Day:2009-09-17_Virtualization To give you an idea of the success of the day, here's a selection of the bugs reported: https://bugzilla.redhat.com/523953 (mbanas) libvirtd segfault with NIC hot-unplug https://bugzilla.redhat.com/524012 (vbenes) libvirt cannot hot-unplug devices which were not previously hot-plugged https://bugzilla.redhat.com/524022 (vbenes) qemu gets confused with NIC hotplug when no model is specified https://bugzilla.redhat.com/523914 (jdenemar) Mouse does not move in PV Xen guest under RHEL-5.4 https://bugzilla.redhat.com/523941 (jdenemar) kernel 2.6.31-1[24].fc12 doesn't boot in xen PV guest https://bugzilla.redhat.com/523971 (jdenemar) xm save hangs with kernel-2.6.31-14.fc12 running as a PV guest under RHEL-5.4 https://bugzilla.redhat.com/523939 (mrezanin) Save on restored machine failed https://bugzilla.redhat.com/523623 (drjones) grub.conf needs console=hvc0 in kernel command-line when installed as DomU https://bugzilla.redhat.com/524039 (drjones) block device cannot be detached from DomU https://bugzilla.redhat.com/524052 (drjones) Boot hang when installing HVM DomU https://bugzilla.redhat.com/524035 (vbenes) libvirt should support USB device detach https://bugzilla.redhat.com/524045 (vbenes) accessing non existing usb device cause guest to stop https://bugzilla.redhat.com/523158 (santwana) libvirtd segault during virsh restore https://bugzilla.redhat.com/523956 (clalance) Starting libvirtd by hand causes denials https://bugzilla.redhat.com/523970 (jstodola) virsh help output not sorted https://bugzilla.redhat.com/523993 (clalance) KVM Live migration failure with SELinux Enforcing https://bugzilla.redhat.com/524047 (gianluca) virt-viewer outputs nothing if connecting to a non existent VM https://bugzilla.redhat.com/522683 (plambert) USB devices do not mount and are not seen by 12-alpha KVM https://bugzilla.redhat.com/523948 (phan) VM cannot boot from the disk converted by qemu-img https://bugzilla.redhat.com/524229 (santwana) Local migration of kvm guest fails in Fedora12 Alpha https://bugzilla.redhat.com/524268 (santwana) KVM guest fails to start up after virt-snapshot in Fedora12 https://bugzilla.redhat.com/524033 (jstodala) libvirt should check the permissions on all paths before starting qemu https://bugzilla.redhat.com/523960 (mnowak) virtinst saves images where qemu can't access them by default https://bugzilla.redhat.com/517379 (adamw) virt-manager should warn if guest images will are not readable by qemu https://bugzilla.redhat.com/524083 (emcnabb) virt-manager storage "Allocation" field can be set higher than "Max Capacity" https://bugzilla.redhat.com/524085 (emcnabb) Help button on virt-manager "New Storage Volume" page broken https://bugzilla.redhat.com/524109 (lmr) virt-manager: Fails to install guest using ISO file - internal error unable to start guest: qemu: 'iso' invalid format https://bugzilla.redhat.com/524111 (lmr) virt-manager can't hold pointer grab on VMs (VNC mode) https://bugzilla.redhat.com/524117 (lmr) virt-manager: Error installing guests - can't mount the / filesystem after the install process is finished https://bugzilla.redhat.com/524118 (P Rauser) NFS4 connection to virtual guest NFS4 server fails over bridged interface https://bugzilla.redhat.com/524205 (vbenes) virt-manager cloning operations should be cancelable Merging Lists ============= Jon Stanley raised the issue again of whether the fedora-virt and fedora-xen mailing lists should be merged: http://www.redhat.com/archives/fedora-virt/2009-October/thread.html#00000 It looks like we'll do this when the lists switch over to the lists.fedoraproject.org server. virt-dostuff ============ Rich Jones announce another couple of tools in the ever growing libguestfs toolsuite, virt-rescue and virt-edit: http://www.redhat.com/archives/fedora-virt/2009-September/msg00099.html Bugs ==== DOOM-O-METER: 198 bugs open now, 214 open three weeks ago. Progress! The Fedora 12 blocker list is: https://bugzilla.redhat.com/showdependencytree.cgi?id=F12VirtBlocker&hide_r… and the Fedora 12 target list is: https://bugzilla.redhat.com/showdependencytree.cgi?id=F12VirtTarget&hide_re… All these queries can be found on: https://fedoraproject.org/wiki/Virtualization_bugs Ongoing Bugs ============ == kernel == https://bugzilla.redhat.com/526869 Add virtio_blk support cache flush (VIRTIO_BLK_F_FLUSH) Christoph points out that we should make sure F-12 virtio_blk supports cache flushing for future host versions. https://bugzilla.redhat.com/525890 Backport virtio patches for optimised virtio-net Some debate about whether backporting some changes from 2.6.32 virtio_net is worth the risk. https://bugzilla.redhat.com/524508 kvm regression between 2.6.29 and 2.6.30 causes guest to become unresponsive Reporter has confirmed that his KVM hangs are a regression between 2.6.29 and 2.6.30. https://bugzilla.redhat.com/523900 Garbage in KVM guest console when guest runs kernel 2.6.30.5-43.fc11 A video console under KVM regression between 2.6.29 and 2.6.30. https://bugzilla.redhat.com/524808 kernel-2.6.31-33.fc12.x86_64 fail to boot with VT-d enabled and intel_iommu=on https://bugzilla.redhat.com/490477 [Intel IOMMU] Using isochronous DMAR unit on ICH10 board causes _other_ DMAR unit to stop working. David Woodhouse continues fighting away at VT-d bugs. == misc == https://bugzilla.redhat.com/501720 upstart does not launch a login process on /dev/hvc0 in all cases Since virtio_console is undergoing major changes, we've punted this until Fedora 13. https://bugzilla.redhat.com/492082 Re-phrase anaconda's terrifying "uninitialized drive" warning Looks like anaconda devs just want to close this. https://bugzilla.redhat.com/523296 When Qemu/SDL is fullscreen xrandr dual-monitor configuration breaks Looks like SDL's fullscreen support can somehow screw up xrandr configuration. == qemu == https://bugzilla.redhat.com/524734 KVM guest ext3 errors at shutdown when using virtio and a qcow2 backing file An Ubuntu guest on F12 using a qcow2 backing file and virtio sees ext3 errors. https://bugzilla.redhat.com/526381 qcow2 performance bad under i/o load Quite worrying given tha qcow2 performance is an F-12 feature. However, it looks like this may be just that these are compressed qcow2 which are not expected to have good performance. https://bugzilla.redhat.com/526777 Guest PXE booting doesn't work when using ne2k_pci NIC model https://bugzilla.redhat.com/526776 Add rtl8209 to gpxe-roms-qemu Eduardo points out that we're including the wrong PXE ROM for ne2k_pci. This is also true in F-11. https://bugzilla.redhat.com/523312 KVM guest freezes during DesktopBSD 1.7 installation This issue on a Fedora 11 host sounds like it might be storage related. https://bugzilla.redhat.com/523592 qemu-kvm segfault when attaching USB audio device We have a stack trace for this main loop segfault on F11 with USB passthrough. https://bugzilla.redhat.com/524723 Windows XP KVM gets BSOD with HP all-in-one usb device attached An issue with USB passthrough on Fedora 11. https://bugzilla.redhat.com/503156 qemu VNC :: xterm inside VM shows garbled text A patch destined for 0.10.7 that should have fixed this has, in fact, made things worse. == libvirt == https://bugzilla.redhat.com/524732 org.libvirt.manage policy kit denial Tom Horsley is seeing polkit denials with virsh and virt-maanger. https://bugzilla.redhat.com/517157 libvirt fails to start guest on NFS even when sebool virt_use_nfs is on F12 has a fix to handle EOPNOTSUPP from setfilecon() when running on NFS. Need this in F11 too. https://bugzilla.redhat.com/516430 libvirt cannot re-label a disk image under an NTFS partition Looks like a similar issue to the NFS one, except we don't have a virt_use_ntfs sebool. https://bugzilla.redhat.com/524770 'virsh vol-path' command doesn't support "name" parameter https://bugzilla.redhat.com/524776 Describe vol-key vol-path, vol-name commands wrongly in help doc Some of the storage volume commands seem to be quite messed up ever since they were introduced. https://bugzilla.redhat.com/525073 cannot delete storage pool with virsh pool-delete Another issue with virsh's storage commands. https://bugzilla.redhat.com/526769 libvirt logrotate settings should roll over weekly instead of daily libvirt's logrotate settings are proving to be quite annoying, so we're switching it to only roll-over weekly. == virt-manager == https://bugzilla.redhat.com/522823 virt-manager doc screenshots are outdated The Fedora 12 version of virt-manager has a radically different UI, but the help documentation hasn't been updated yet. https://bugzilla.redhat.com/522720 virt-manager cloning a guest with a 3Gib image creates a 23Gib image virt-manager appears to get confused about image sizes when cloning a guest. https://bugzilla.redhat.com/524297 F11 virt-manager doesn't allow KVM memory ballooning Looks like this is fixed in F12, so it's purely a case of whether we should make the effort to fix in F11 too. https://bugzilla.redhat.com/526077 virt-manager toolbar buttons should have tooltips According to the GNOME HIG, all toolbar buttons should have tooltips, especially ones without a label. https://bugzilla.redhat.com/526076 some virt-manager toolbar icons are blurry Eagle eyed Michael Monreal spotted that some of virt-manager's new icons seem to be blurry due to inappropriate scaling. He even included a nice screenshot demonstrating the problem. https://bugzilla.redhat.com/526111 virt-manager add storage volume allows an empty name field Minor problem with virt-manager not checking the volume name text entry for data. https://bugzilla.redhat.com/526488 virt-manager's 'Details' dialog has a poorly placed 'Remove' button https://bugzilla.redhat.com/526491 virt-manager has no confirmation dialog when deleting a virtual disk Matt Booth accidentally deleted a few guest images because of a combination of these bugs. https://bugzilla.redhat.com/526648 virt-maanger's "New VM" window disappears when connection expander is closed Yet another minor virt-manager UI quirklet. == virt-clone == https://bugzilla.redhat.com/524269 virt-clone should demove old udev rules when changing MAC address Could be implemented using libguestfs. == xen == https://bugzilla.redhat.com/523914 Mouse does not move in PV Xen guest under RHEL-5.4 A rather interesting, complex, twisty bug report detailing why xen and evdev have conspired to break the mouse in Fedora 12 Xen guests. https://bugzilla.redhat.com/523489 32-bit rawhide xen kernel spins 100% cpu booting on CentOS 5.2 Dom0 Xen DomU boot failure. Resolved Bugs ============= == misc == https://bugzilla.redhat.com/507397 Directory permissions on volume group directory too restrictive This blocker issue with lvm2 was first fixed by a large change to using udev, but that broken anaconda so it was reverted, and then it was fixed by a much more minor change. == kernel == https://bugzilla.redhat.com/520119 virtio_net page allocation failure Rusty came up with a fix for this in 2.6.31, so we need to backport it to 2.6.30 in Fedora 11. https://bugzilla.redhat.com/524229 Local migration of kvm guest fails in Fedora12 Alpha A PVMMU bug which has been fixed in 2.6.31. https://bugzilla.redhat.com/509383 rotational mode is much faster for virtio-blk disks, but uses non-rotational mode by default Justin has applied Christoph's fix for this in rawhide and F-11. https://bugzilla.redhat.com/512358 Unable to boot using qemu-kvm and gPXE from virt-preview repository This is now fixed in Fedora 11. == qemu == https://bugzilla.redhat.com/524022 qemu's pci_add monitor command should not exit() if the NIC model is not valid Markus fixed this upstream and the patches have now been pulled into F12. https://bugzilla.redhat.com/524695 qemu should print strerror info for disk/kernel/initrd access errors Cole points out that since we're getting a lot of problems with qemu not being able to access files since we made it run as an unprivileged user, we should really fix qemu's error messages. Justin has sent a patch upstream and included it for F12. https://bugzilla.redhat.com/522683 svirt blocks USB passthrough, even with virt_use_usb enabled - /sys/bus/usb/devices Looks like svirt is breaking USB passthrough by blocking access to /sys/bus/usb/devices. Dan Walsh fixed this in rawhide. https://bugzilla.redhat.com/522994 KVM Fedora 11 guest networking fails with latest (2.6.30.5-43) kernel. F-11 qemu-kvm was missing a crucial piece of GSO support which caused networking to break with 2.6.30 guests. This was only a problem for people using qemu-kvm directly from the command line because libvirt wasn't enabling GSO (#526472). Fixed now in updates-testing. https://bugzilla.redhat.com/509532 Move /usr/bin/qemu-kvm into the qemu-kvm package danpb points out that the qemu-kvm package should go away eventually, so we should keep it as an empty meta-package for comps. https://bugzilla.redhat.com/523677 add support for s390x Dan Horák proposed that Ulrich Hecht's patch to add support for s390 to qemu be included in Fedora, but it was decided to just wait for the patch to get upstream. == libvirt == https://bugzilla.redhat.com/524499 libvirt should not enable qemu's audio backend, even with selinux disabled Now that we run qemu as an unprivileged user, we should never enable qemu's sdl/pulse audio backend. A patch to do that from upstream is now in F-12. https://bugzilla.redhat.com/497131 svirt fails to relabel qcow2 backing files This has been fixed by using libvirt's existing image format probing code to determine the backing file and then re-labelling it. Patch has been backported to F-12 now. https://bugzilla.redhat.com/526472 libvirt fails to enable IFF_VNET_HDR on tap fds for qemu-kvm In F-11 libvirt was failing to recognise that qemu had GSO support; fixed by backporting patches to re-work how libvirt probes qemu versions. https://bugzilla.redhat.com/523766 AVCs from libvirtd startup A bunch of SELinux AVCs during libvirtd startup that look like they might be fixed by latest upstream netcf. Fixed by Dan Walsh in latest rawhide. https://bugzilla.redhat.com/517617 libvirt/netcf loads modprobe.conf and others - AVC messages (preventing libvirtd (virtd_t) "getattr" modules_conf_t) More netcf related AVCs which have been fixed. https://bugzilla.redhat.com/520864 libvirt is using untrusted 'info vcpus' PID data for already running VM after libvirtd restart https://bugzilla.redhat.com/465532 RFE: libvirt should support KVM huge page backed memory https://bugzilla.redhat.com/517619 libvirtd should chown dirs when qemu configured to run as root/root All fixed in rawhide by libvirt-0.7.1. == python-virtinst == https://bugzilla.redhat.com/523960 virtinst saves images where qemu can't access them by default Cole has fixed this by making virtinst added qemu to the ACLs for the directories qemu will need access to. https://bugzilla.redhat.com/524109 virt-manager: Fails to install guest using ISO file - internal error unable to start guest: qemu: 'iso' invalid format The fix for bug #517151 introduces another problem; Cole has now fixed that one in rawhide. https://bugzilla.redhat.com/499654 failure to add more than 16 virtio-blk devices in virt-manager https://bugzilla.redhat.com/505317 virtinst: make SLES11 guests use virtio by default https://bugzilla.redhat.com/506319 virtinst errors finding default bridge: upsets virt-manager Various F11 virtinst bugs fixed by python-virtinst-0.400.3-9.fc11. == virt-manager == https://bugzilla.redhat.com/517379 virt-manager should warn if guest images will are not readable by qemu When attaching any storage to a guest we now attempt to verify that the 'qemu' user has search access. If not, we warn the user and offer to fix, using ACLs. The user can opt out, and optionally 'Never ask about this path again" if we have any false positives. https://bugzilla.redhat.com/517664 virt-manager ignores "Host does not support any virtualization options" error Cole has made this error condition a bit more obvious to the user. https://bugzilla.redhat.com/522823 virt-manager doc screenshots are outdated The help docs are so woefully out of date, Cole has just disabled them for now. https://bugzilla.redhat.com/502777 virt-manager does not recalculate free disk space on new VM creation While the 'New VM' dialog is up, virt-manager now polls every 3 seconds to check the available disk space on the host.

2 1

Re: [fedora-virt] virt-preview updates
by Gianluca Cecchi 02 Oct '09

02 Oct '09

On Thu, 01 Oct 2009 21:59:01 +0100 Mark McLoughlin wrote: > == virt-manager == > > * Tue Sep 29 2009 Cole Robinson <crobinso redhat com> - 0.8.0-6.fc12 > - Fix VCPU hotplug Hello, any info on about what would have been fixed? And also provided functionality itself.... In my env with 0.8.0-6 and f11 with virt-preview it seems I cannot hotplug/unplug at all a cpu.... In details window of a rh 5.3 vm with 2 vCPUs, I cannot raise the number (current is equal to maximum), while I can lower but I get that the effect will be visible at next reboot..... Or do I have to create a new VM with this peculiar version of virt-manager to see different behaviour? Gianluca

2 1

merging lists?
by Jon Stanley 01 Oct '09

01 Oct '09

I'm working on migrating Fedora lists from RHT infrastructure to Fedora infrastructure in the coming little bit. The question that I had was does fedora-xen and fedora-virt both need to exist? I'm subscribed to both, and certainly the conversations on each are different, but could they be merged? As of several months ago, the subscriber count of fedora-xen was WAY higher than that of fedora-virt, this doesn't seem to make much sense since Xen is not The Way Forward(TM), at least from what I can see. I guess the question is could we merge the two lists, and have one less to migrate, or are both really needed?

4 4

using kqemu with fedora-virt-preview repo
by Gianluca Cecchi 01 Oct '09

01 Oct '09

I have a powerful server but without VT capabilities, where I would like to use virt-manager tools as of fedora-virt-preview repo and speed up somehow things using kqemu. It seems that kqemu support is not compiled in for this repo, correct? Instead, as the kqemu kernel module is in rpmfusion, I presume that the stock f11 qemu has this support built in.... Am I wrong? Any problems to compile kqemu support ? On another server, tried to rebuild qemu-0.11.0-2.fc11.src.rpm (from fedora-virt) after installing kqemu from rpmfusion. But while if I take the source I get kqemu as enabled in configure script, in rpmbuild I don't get so. I notice that inside the qemu tar.gz as provided by the source.rpm there are three lines below in the top configure script: if [ "$cpu" = "i386" -o "$cpu" = "x86_64" ] ; then kqemu="yes" audio_possible_drivers="$audio_possible_drivers fmod" kvm="yes" kqemu="no" So I commented out the kqemu="no" line... and built the package (not touching the other two configure scripts that seems related to kvm...) After installing, if I run qem-kvm or qemu without kqemu all is ok. Instead, running [root@tekkaman bin]# ./qemu --enable-kqemu I get Not enough memory (requested_size = 16777216, max memory = 146800640) Aborted Any hints? For Kqemu now gplv2, are there any barriers to have it included in fedora directly? Thanks in advance, Gianluca

2 2

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

virt October 2009