Re: [Xen-users] vhd support in xen-4.4

[Ian Campbell <ian.campbell@xxxxxxxxxx>]

So it looks like udev is running this command for some reason, which
sadly doesn't tell us much I'm afraid (I had hoped there would be some
obvious script in the middle).

If you are using Xen4Centos then you might want to try the CentOS Xen
list, you are more likely to find people who have come across this
before.

On Tue, 2015-05-12 at 17:26 +0200, dmitry@xxxxxxxxxx wrote:
> pstree -s
> 
> pstree: invalid option -- 's'
> 
> ps auxf
> 
> root        1  0.0  0.1  19284  1500 ?         Ss  May05     0:00 /sbin/init
> 
> root      478  0.0  0.1  11256  1540 ?        S<s  May05   0:01 /sbin/udevd -d
> 
> root    14898  0.0  0.2  11908  2060 ?       S<  17:16      0:00  \_ 
> /sbin/udevd -d
> 
> root    15047  0.0  0.0  8368  368 ?          D<  17:16      0:00  |  \_ 
> /sbin/blkid -o udev -p /dev/tdf
> 
> root    14900  0.0  0.1  11252  1304 ?       S<  17:16      0:00  \_ 
> /sbin/udevd -d
> 
> root    15053  0.0  0.1  11252  1280 ?       S<  17:16      0:00  \_ 
> /sbin/udevd -d
> 
> root    15140  0.0  0.1  11252  1216 ?       S<  17:17      0:00  \_ 
> /sbin/udevd -d
> 
> 
> 12.05.2015, 16:56, "Ian Campbell" <ian.campbell@xxxxxxxxxx>:
> >  On Tue, 2015-05-12 at 16:45 +0200, dmitry@xxxxxxxxxx wrote:
> >>   This process come only when windows hvm domain try to start.
> >>   First start Windows domain with VHD disk not successful.
> >>   Next I xl destroy this domain, kill process.
> >>   Then I successful start domain.
> >>
> >>   ps auxf | grep " D"
> >>   root     13696  0.0  0.0   8368   368 ?        D<   16:13   0:00  |   \_ 
> >> /sbin/blkid -o udev -p /dev/tdf
> >>   root     13864  0.0  0.0 103296   836 pts/0    S+   16:14   0:00         
> >>              \_ grep  D
> >>
> >>   pstree 13696
> >  I said "pstree -s", which will show you the parents of the process,
> >  which you have cut out from your ps output too.
> >
> >  The parent processes are interesting because that should tells us where
> >  that process is getting launched from.
> >
> >  Ian.
> >>   blkid
> >>
> >>   I installed Xen from CentOS repo (Xen4CentOS).
> >>
> >>   12.05.2015, 10:58, "Ian Campbell" <ian.campbell@xxxxxxxxxx>:
> >>>   On Tue, 2015-05-05 at 09:24 +0200, dmitry@xxxxxxxxxx wrote:
> >>>>    After it I destroy domain (xl destroy).
> >>>>    ps aux | grep " D" show process
> >>>>    root     16443  0.0  0.0   8368   368 ?        D<   08:39   0:00 
> >>>> /sbin/blkid -o udev -p /dev/tdf
> >>>>
> >>>>    I kill it (kill -s 9 16443). If I don't do it, domain does not start.
> >>>   Do you know where this process comes from? AFAIK it is not spawned by
> >>>   anything in the Xen infrastructure. It might be some other facility on
> >>>   your system.
> >>>
> >>>   My hypothesis is that this process, whatever it is, is holding the
> >>>   tapdisk from the previous domain open, preventing it from closing down,
> >>>   which in turn causes the next domain to fail because the backing file is
> >>>   in use.
> >>>
> >>>   pstree -s $PID-OF-BLKID
> >>>
> >>>   might give a clue, or look in "ps auxf" and look at the parent
> >>>   processes.
> >>>
> >>>   Which blktap kernel module are you using? And are you using the tapdisk
> >>>   from xen.git or from e.g. the xenserveer repos?
> >>>
> >>>   Ian.



_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users