Hello everybody,
I think I''ve screwed up today. I was trying to fix a malfunctioning
domU and I ended up issuing xenstore-rm {/local,/vm}.
I feel like an idiot that did something really wrong. Since that, all the my
virtual machines cannot start anymore.
What I get is:
xevelon ~ # xl cr -c /xen/configs/aquaria
Parsing config from /xen/configs/aquaria
libxl: error: libxl_device.c:858:device_backend_callback: unable to disconnect
device with path /local/domain/0/backend/vbd/12/51713
libxl: error: libxl_device.c:858:device_backend_callback: unable to disconnect
device with path /local/domain/0/backend/vbd/12/51714
libxl: error: libxl_device.c:858:device_backend_callback: unable to disconnect
device with path /local/domain/0/backend/vbd/12/51715
libxl: error: libxl_device.c:858:device_backend_callback: unable to disconnect
device with path /local/domain/0/backend/vbd/12/51728
libxl: error: libxl_create.c:933:domcreate_launch_dm: unable to add disk devices
libxl: error: libxl_device.c:858:device_backend_callback: unable to disconnect
device with path /local/domain/0/backend/vbd/12/51713
libxl: error: libxl_device.c:858:device_backend_callback: unable to disconnect
device with path /local/domain/0/backend/vbd/12/51714
libxl: error: libxl_device.c:858:device_backend_callback: unable to disconnect
device with path /local/domain/0/backend/vbd/12/51715
libxl: error: libxl_device.c:858:device_backend_callback: unable to disconnect
device with path /local/domain/0/backend/vbd/12/51728
libxl: error: libxl.c:1465:devices_destroy_cb: libxl__devices_destroy failed for
12
And if I take a look at xenstored:
xevelon ~ # xenstore-ls -fp
/tool = "" (n0)
/tool/xenstored = "" (n0)
/libxl = "" (n0)
/vm = "" (n0)
/local = "" (n0)
/local/domain = "" (n0)
/local/domain/0 = "" (n0)
/local/domain/0/name = "Domain-0" (n0)
/local/domain/0/libxl = "" (n0)
/local/domain/0/libxl/disable_udev = "1" (n0)
Obviously something is wrong here. I''ve no data about the dom0, no
information about the backend. Is there a way to reload all the stuff
I''ve removed?
Thank you very much
--
Fabiano Francesconi [GPG key: 0x81E53461]
This seems like a -users question to me, please try and pick the appropriate list for your questions. I''ve moved this there. On Mon, 2013-02-11 at 14:53 +0000, Fabiano Francesconi wrote:> Is there a way to reload all the stuff I''ve removed?Rebooting is the most obvious way to reset to a known good state quickly. You could also try restarting the "xencommons" service. However the only thing which should be in xenstore at start of day is the domain 0 name, the rest should be created on the fly as needed, and what you have now looks a lot like a fresh system to me, so I''m not quite sure what is causing your issue. I''m not 100% sure but perhaps this has also broken the backend driver''s watches on the xenstore. But I thought they were path based, rather than based on the specific node, and so shouldn''t be affected but perhaps I am mistaken. Anyway, you could try rmmod and modprobe all of the Xen related modules in your kernel (assuming they are mostly modular). Ian.
This seems like a -users question to me, please try and pick the appropriate list for your questions. I''ve moved this there. On Mon, 2013-02-11 at 14:53 +0000, Fabiano Francesconi wrote:> Is there a way to reload all the stuff I''ve removed?Rebooting is the most obvious way to reset to a known good state quickly. You could also try restarting the "xencommons" service. However the only thing which should be in xenstore at start of day is the domain 0 name, the rest should be created on the fly as needed, and what you have now looks a lot like a fresh system to me, so I''m not quite sure what is causing your issue. I''m not 100% sure but perhaps this has also broken the backend driver''s watches on the xenstore. But I thought they were path based, rather than based on the specific node, and so shouldn''t be affected but perhaps I am mistaken. Anyway, you could try rmmod and modprobe all of the Xen related modules in your kernel (assuming they are mostly modular). Ian.
Fabiano Francesconi
2013-Feb-11 15:12 UTC
Re: [Xen-devel] Xen 4.2.0: unable to add disk devices
On Monday 11 February 2013 at 16:06, Ian Campbell wrote:> This seems like a -users question to me, please try and pick the > appropriate list for your questions. I''ve moved this there.Sorry and thank you!> > On Mon, 2013-02-11 at 14:53 +0000, Fabiano Francesconi wrote: > > Is there a way to reload all the stuff I''ve removed? > > > > Rebooting is the most obvious way to reset to a known good state > quickly. > > You could also try restarting the "xencommons" service. > > However the only thing which should be in xenstore at start of day is > the domain 0 name, the rest should be created on the fly as needed, and > what you have now looks a lot like a fresh system to me, so I''m not > quite sure what is causing your issue. > > I''m not 100% sure but perhaps this has also broken the backend driver''s > watches on the xenstore. But I thought they were path based, rather than > based on the specific node, and so shouldn''t be affected but perhaps I > am mistaken. Anyway, you could try rmmod and modprobe all of the Xen > related modules in your kernel (assuming they are mostly modular). > > Ian.Rebooting solved the issue. I tried rebooting twice, before, but it didn''t help. This time I completely turned off the server and turned it back on. Sorry but it is the whole day that I''m working on these machines and I hurried up writing here while I should have spent more time digging into it by myself. Still, thank you very much. -- Fabiano Francesconi