Bug#650711: [Pkg-libvirt-maintainers] Bug#650711: virt-manager gets stuck
forcemerge 663931 650711 thanks On Mon, Jun 11, 2012 at 11:30:48AM +0200, Harald Dunkel wrote: > Hi Guido, > > On 06/08/12 16:00, Guido Günther wrote: > > > > Could that be the same as 663931? I've attached a possible workaround. > > Cheers, > > -- Guido > > > > I haven't tried your patch (yet), but restarting udev on > the libvirt server seems to help. Before the restart virt-manager > got stuck trying to connect the server. After the restart it > worked. I tried this on 3 libvirt servers. Looks pretty much > like #663931. Thanks a lot for testing! > > Unfortunately I won't be able to patch and restart libvirtd > on these servers. They are in production. No problem. I intend to roll the patch in one of the next releases (in case we don't find a better fix). I'd be great if you could report back then. Cheers, -- Guido -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#650711: [Pkg-libvirt-maintainers] Bug#650711: virt-manager gets stuck
Hi Guido, On 06/08/12 16:00, Guido Günther wrote: > > Could that be the same as 663931? I've attached a possible workaround. > Cheers, > -- Guido > I haven't tried your patch (yet), but restarting udev on the libvirt server seems to help. Before the restart virt-manager got stuck trying to connect the server. After the restart it worked. I tried this on 3 libvirt servers. Looks pretty much like #663931. Unfortunately I won't be able to patch and restart libvirtd on these servers. They are in production. Regards Harri -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#650711: [Pkg-libvirt-maintainers] Bug#650711: virt-manager gets stuck
On Mon, Jan 02, 2012 at 04:40:09PM +0100, Harald Dunkel wrote: > On 01/02/12 15:58, Guido Günther wrote: > > > > Thanks for reporting back! > > > >> > >> This is virt-manager version 0.9.0-3~bpo60+1 and > >> libvirt0 version 0.9.8-2~bpo60+2. > > > > These versions were never uploaded, is this self compiled? Could you > > check with the versions in sid/wheezy? > > These are the versions from Wheezy. I just built them for > Squeeze. To install virt-manager on Squeeze I had to move > the spice gtk GUI to Recommends, the rest is unchanged AFAIR. > > Unfortunately I do not have the option to update my > KVM servers to Wheezy. > > > Are the remote libvirt versions also based on 0.9.8 or are these older > > ones? Can you strace/debug the remote libvirt daemons to check where the > > broken pipe comes from? > > Using "localhost" the libvirt daemon is not remote anymore. > All machines are running Squeeze with the given versions. > > Attached you can find the strace output. This is a live > system, so I have attached strace to libvirtd using > "strace -p". Hopefully this is OK? > > I started strace, then I started virt-manager on the same > host and tried to connect to localhost. When it came back > with "Not connected" I interrupted strace with ^C. > > > Hope this helps. Please mail. Could that be the same as 663931? I've attached a possible workaround. Cheers, -- Guido -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#650711: Info received ([Pkg-libvirt-maintainers] Bug#650711: virt-manager gets stuck)
I found some more information in the libvirtd.log file. When virt-manager becomes responsive to show localhost (QEMU) - Not connected again, then libvirtd writes warning : virKeepAliveTimer:182 : No response from client 0x7ffb4c0d9290 after 5 keepalive messages in 30 seconds into it's log file. Hope this helps. Regards Harri -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#650711: [Pkg-libvirt-maintainers] Bug#650711: virt-manager gets stuck
On Mon, Jan 02, 2012 at 02:11:11PM +0100, Harald Dunkel wrote: > Attached you can find the output of "virt-manager --debug". > I tried to connect to localhost using qemu:///system. > virt-manager got stuck for a few minutes after the xml > file, then it said "Not connected". "virsh list --all" > does not get stuck. Thanks for reporting back! > > This is virt-manager version 0.9.0-3~bpo60+1 and > libvirt0 version 0.9.8-2~bpo60+2. These versions were never uploaded, is this self compiled? Could you check with the versions in sid/wheezy? Are the remote libvirt versions also based on 0.9.8 or are these older ones? Can you strace/debug the remote libvirt daemons to check where the broken pipe comes from? Cheers, -- Guido -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#650711: virt-manager gets stuck
Attached you can find the output of "virt-manager --debug".
I tried to connect to localhost using qemu:///system.
virt-manager got stuck for a few minutes after the xml
file, then it said "Not connected". "virsh list --all"
does not get stuck.
This is virt-manager version 0.9.0-3~bpo60+1 and
libvirt0 version 0.9.8-2~bpo60+2.
Please mail if I can help to track this down.
Regards
Harri
2012-01-02 14:02:29,431 (cli:71): virt-manager startup
2012-01-02 14:02:29,432 (virt-manager:292): Launched as:
/usr/share/virt-manager/virt-manager.py --debug
2012-01-02 14:02:29,432 (virt-manager:293): GTK version: (2, 20, 1)
2012-01-02 14:02:29,432 (virt-manager:294): virtManager import:
2012-01-02 14:02:29,515 (keyring:30): gnomekeyring bindings not installed, no
keyring support
2012-01-02 14:02:29,556 (engine:555): No inspection thread because libguestfs
is too old, not available, or libvirt is not thread safe.
2012-01-02 14:02:29,558 (engine:346): About to connect to uris
['qemu+ssh://hdunkel@srvl022/system', 'qemu+ssh://hdunkel@srvl024/system',
'qemu+ssh://hdunkel@dpcl082/system', 'qemu+ssh://hdunkel@srvl023/system',
'qemu+ssh://hdunkel@srvl021/system', 'qemu+ssh://hdunkel@srvl026/system',
'qemu+ssh://hdunkel@srvl025/system', 'qemu+ssh://hdunkel@srvl014/system',
'qemu:///system']
2012-01-02 14:02:29,632 (engine:471): window counter incremented to 1
2012-01-02 14:02:38,481 (connection:954): Scheduling background open thread for
qemu:///system
2012-01-02 14:02:38,482 (connection:1140): Background 'open connection' thread
is running
2012-01-02 14:02:38,486 (connection:1168): Background open thread complete,
scheduling notify
2012-01-02 14:02:38,487 (connection:1173): Notifying open result
2012-01-02 14:02:38,516 (connection:1180): qemu:///system capabilities:
628110e3-0202-11df-ad03-000bab025014
x86_64
Nehalem
Intel
tcp
hvm
32
/usr/bin/kvm
pc-0.12
pc
pc-0.11
pc-0.10
isapc
/usr/bin/kvm
hvm
64
/usr/bin/kvm
pc-0.12
pc
pc-0.11
pc-0.10
isapc
/usr/bin/kvm
2012-01-02 14:03:08,549 (util:62): Attempting to build default pool with target
'/var/lib/libvirt/images'
2012-01-02 14:03:08,549 (connection:493): Building default pool failed:
Couldn't create default storage pool '/var/lib/libvirt/images': Cannot write
data: Broken pipe
2012-01-02 14:03:08,550 (connection:1215): Unable to list active pools: Cannot
write data: Broken pipe
2012-01-02 14:03:08,550 (connection:1219): Unable to list inactive pools:
Cannot write data: Broken pipe
2012-01-02 14:03:08,553 (cli:85): Traceback (most recent call last):
File "/usr/share/virt-manager/virtManager/baseclass.py", line 33, in
_safe_wrapper
return func(*args)
File "/usr/share/virt-manager/virtManager/connection.py", line 1182, in
_open_notify
self.tick()
File "/usr/share/virt-manager/virtManager/connection.py", line 1422, in tick
newInterfaces, self.interfaces) = self._update_interfaces()
File "/usr/share/virt-manager/virtManager/connection.py", line 1303, in
_update_interfaces
lookup_func, build_class)
File "/usr/share/virt-manager/virtManager/connection.py", line 1209, in
_poll_helper
if not check_support():
File "/usr/share/virt-manager/virtManager/connection.py", line 512, in
is_interface_capable
virtinst.support.SUPPORT_CONN_INTERFACE)
File "/usr/lib/pymodules/python2.6/virtinst/support.py", line 574, in
check_conn_support
return _check_support(conn, feature, conn)
File "/usr/lib/pymodules/python2.6/virtinst/support.py", line 443, in
_check_support
actual_drv_ver = _hv_ver(conn, uri)
File "/usr/lib/pymodules/python2.6/virtinst/support.py", line 376, in _hv_ver
ret = cmd(*args)
File "/usr/lib/python2.6/dist-packages/libvirt.py", line 2884, in getVersion
if ret == -1: raise libvirtError ('virConnectGetVersion() failed',
conn=self)
libvirtError: Cannot write data: Broken pipe
None
Traceback (most recent call last):
File "/usr/share/virt-manager/virtManager/baseclass.py", line 33, in
_safe_wrapper
return func(*args)
File "/usr/share/virt-manager/virtManager/connection.py", line 1182, in
_open_notify
self.tick()
File "/usr/share/virt-manager/virtManager/connection.py", line 1422, in tick
newInterfaces, s
Bug#650711: virt-manager gets stuck
I doubt that the bug report at RedHat is about the same problem. In my problem virt-manager doesn't "freeze frequently" or "from time to time", but at start up time. It is reproducible. Regards Harri -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#650711: [Pkg-libvirt-maintainers] Bug#650711: virt-manager gets stuck
On Fri, Dec 02, 2011 at 09:20:29AM +0100, Harald Dunkel wrote: > Package: virt-manager > Version: 0.9.0-3 > > I have 7 remote hosts connected to virt-manager (using ssh). > Problem: At start time virt-manager gets stuck for at least > 20 minutes. During this time the GUI is completely unresponsive > and doesn't refresh. > > I see the ssh sessions in a local "ps -ef" while it is stuck: > > % ps -ef | grep virt > root 2382 1 0 Nov27 ?00:00:00 /usr/sbin/libvirtd -d > hdunkel 10411 1 0 15:58 ?00:00:00 python > /usr/share/virt-manager/virt-manager.py > hdunkel 10415 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl025 sh -c > 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then > ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' > hdunkel 10418 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl026 sh -c > 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then > ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' > hdunkel 10419 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl022 sh -c > 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then > ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' > hdunkel 10422 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl024 sh -c > 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then > ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' > hdunkel 10424 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl021 sh -c > 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then > ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' > hdunkel 10425 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl014 sh -c > 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then > ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' > hdunkel 10427 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl023 sh -c > 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then > ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' > hdunkel 10429 10119 0 15:58 pts/000:00:00 grep virt > > > On the remote side I see "nc -q0 -U /var/run/libvirt/libvirt-sock" > running, as expected. /bin/nc is netcat-openbsd. > > The socket is readable and writable on all remote hosts: > > % ls -al /var/run/libvirt/libvirt-sock > srwxrwx--- 1 root libvirt 0 Nov 27 11:02 /var/run/libvirt/libvirt-sock > > The account I use for ssh is member of the libvirt group. > > > This is a highly annoying problem, because I usually don't > touch the virtual machines, unless there is an urgent issue. > Having to wait for an unresponsive GUI when its urgent is > extremely frustrating. > > The virt-manager in Squeeze shows the same problem. > > Any workaround or helpful comment would be highly appreciated. The first thing would be http://wiki.debian.org/libvirt#Bugreports.2BAC8-Debugging This might indicate a cause for the hang. Cheers, -- Guido -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#650711: virt-manager gets stuck
Package: virt-manager Version: 0.9.0-3 I have 7 remote hosts connected to virt-manager (using ssh). Problem: At start time virt-manager gets stuck for at least 20 minutes. During this time the GUI is completely unresponsive and doesn't refresh. I see the ssh sessions in a local "ps -ef" while it is stuck: % ps -ef | grep virt root 2382 1 0 Nov27 ?00:00:00 /usr/sbin/libvirtd -d hdunkel 10411 1 0 15:58 ?00:00:00 python /usr/share/virt-manager/virt-manager.py hdunkel 10415 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl025 sh -c 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' hdunkel 10418 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl026 sh -c 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' hdunkel 10419 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl022 sh -c 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' hdunkel 10422 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl024 sh -c 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' hdunkel 10424 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl021 sh -c 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' hdunkel 10425 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl014 sh -c 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' hdunkel 10427 10411 0 15:58 ?00:00:00 ssh -l hdunkel srvl023 sh -c 'if nc -q 2>&1 | grep "requires an argument" >/dev/null 2>&1; then ARG=-q0;fi;nc $ARG -U /var/run/libvirt/libvirt-sock' hdunkel 10429 10119 0 15:58 pts/000:00:00 grep virt On the remote side I see "nc -q0 -U /var/run/libvirt/libvirt-sock" running, as expected. /bin/nc is netcat-openbsd. The socket is readable and writable on all remote hosts: % ls -al /var/run/libvirt/libvirt-sock srwxrwx--- 1 root libvirt 0 Nov 27 11:02 /var/run/libvirt/libvirt-sock The account I use for ssh is member of the libvirt group. This is a highly annoying problem, because I usually don't touch the virtual machines, unless there is an urgent issue. Having to wait for an unresponsive GUI when its urgent is extremely frustrating. The virt-manager in Squeeze shows the same problem. Any workaround or helpful comment would be highly appreciated. Regards Harri -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
