[lxc-users] LXC Service fails to start in Fedora

CDR venefax at gmail.com
Thu May 1 17:48:24 UTC 2014


I can confirm that although there is no error message, it does
nothing, I mean, it fails to start any containers.
In my particular case, I need the libvirt bridge virbr0 to be up,
since my container uses this bridge.
Any ideas how fix this without waiting for a patch? I am sure there
must be a way to delay the service start until the bridge is up.


On Thu, May 1, 2014 at 12:29 PM, Michael H. Warfield <mhw at wittsend.com> wrote:
> On Thu, 2014-05-01 at 12:18 -0400, Dwight Engen wrote:
>> On Thu, 01 May 2014 11:50:55 -0400
>> "Michael H. Warfield" <mhw at WittsEnd.com> wrote:
>>
>> > On Thu, 2014-05-01 at 11:44 -0400, Dwight Engen wrote:
>> > > On Thu, 01 May 2014 11:30:02 -0400
>> > > "Michael H. Warfield" <mhw at WittsEnd.com> wrote:
>> > >
>> > > > On Thu, 2014-05-01 at 11:25 -0400, Stéphane Graber wrote:
>> > > > > On Thu, May 01, 2014 at 11:19:28AM -0400, Michael H. Warfield
>> > > > > wrote:
>> > > > > > On Thu, 2014-05-01 at 10:13 -0400, CDR wrote:
>> > > > > > > Actually, /usr/libexec/lxc/lxc-startup is a missing object,
>> > > > > > > not a directory.
>> > > > > > >
>> > > > > > > It seems that the code does not include the two objects,
>> > > > > > > which are the basis for the LXC service.
>> > > > > > >
>> > > > > > > Any idea how can this be fixed?
>> > > > > > >
>> > > > > > > lxc-stratup
>> > > > > >
>> > > > > > You mean lxc-startup.  Yeah that seems to be missing.  I
>> > > > > > think we man have forgotten to glue a couple of pieces
>> > > > > > together.  Stéphane did the lxc-autostart stuff while I was
>> > > > > > working on the systemd service stuff. lxc-startup was a
>> > > > > > temporary shim until lxc-autostart became available. It
>> > > > > > wasn't intended to reach release.  Maybe we each thought the
>> > > > > > other was going to complete that connection or maybe we just
>> > > > > > weren't thinking about it at all.
>> > > > > >
>> > > > > > As a first suggestion, edit the systemd service file:
>> > > > > >
>> > > > > > /usr/lib/systemd/system/lxc.service
>> > > > > >
>> > > > > > Change these lines:
>> > > > > >
>> > > > > > ExecStart=/usr/libexec/lxc/lxc-startup start
>> > > > > > ExecStop=/usr/libexec/lxc/lxc-startup stop
>> > > > > >
>> > > > > > To this:
>> > > > > >
>> > > > > > ExecStart=/usr/bin/lxc-autostart -a
>> > > > > > ExecStop=/usr/bin/lxc-autostart -s
>> > > >
>> > > > > Care to send a patch for that? :)
>> > > >
>> > > > That must be a retorical question.
>> > > >
>> > > > As soon as I can quickly get it tested.  I know you're on a
>> > > > deadline for 1.0.4.  I'm on it.
>> >
>> > > Hey guys, that will kinda work, but if the auto-started containers
>> > > are relying on the bridge setup by libvirt, they can get started
>> > > before libvirt has brought up the bridge even if we make this unit
>> > > depend on libvirtd.service because that service returns to systemd
>> > > before the bridge has been brought up.
>> >
>> > > What I've done is to call the sysvinit script from ExecStart= and
>> > > ExecStop=, and that avoids the problem for me since the sysvinit
>> > > script waits for the bridge.
>> > > libvirt-guests.service/libvirt-guests.sh is a similar arrangement
>> > > where the same script is used as a sysvinit script and is called by
>> > > the .service file. If you want I can make lxc's .service file do
>> > > similar.
>> >
>> > That would be an acceptable alternative but doesn't that run into the
>> > admonishment not to have the sysvinit script present when there's a
>> > systemd service present?  I seem to recall something about that.
>>
>> Yes, that is correct. We don't want to install it as a sysvinit script
>> per se, it will be put into @libexec@/lxc with a different name so it
>> will only be run from the service file, not by any sysvinit
>> compatibility stuff. The contents are the same, it will work when
>> called from the service file on a systemd system or by sysvinit on a
>> traditional system.
>
> Ok...  Sounds reasonable to me.  Especially since I screwed up my first
> attempt at the patch for the above.
>
>> > I just sent in a patch for the change I proposed.  If you feel this is
>> > better and their isn't a conflict, I'm fine with canceling my patch
>> > and going with your suggestion.
>>
>> Well I'm testing it now, when that is done I'll send a patch and you
>> can see what you think.
>>
>> > That order dependency could be also handled in the systemd
>> > dependencies. I hadn't looked at them before.
>>
>> I tried that, but as explained above it didn't work because libvirt
>> returned to systemd before bringing up the bridge, and thus systemd
>> then starts the lxc service thinking that the prerequisite is already
>> done.
>>
>> > > > > > Then try starting.
>> > > > > >
>> > > > > > Let me know if that works.
>> > > > > >
>> > > > > > I guess I didn't spot that in my environment because I have
>> > > > > > the shim present.  :-P
>> > > > > >
>> > > > > > > lxc-devsetup
>> > > > > >
>> > > > > > From what I see below, lxc-devsetup was there.
>> > > > > >
>> > > > > >
>> > > > > > > On Thu, May 1, 2014 at 10:05 AM, CDR <venefax at gmail.com>
>> > > > > > > wrote: I went ahead and created the directory, but I think
>> > > > > > > some code is missing. By the way, this is compiled source
>> > > > > > > code. I did create the RPMs and upgraded the distribution's
>> > > > > > > RPM's
>> > > > > > >         systemctl status lxc.service
>> > > > > > >         lxc.service - LXC Container Initialization and
>> > > > > > > Autoboot Code Loaded: loaded
>> > > > > > > (/usr/lib/systemd/system/lxc.service; enabled)
>> > > > > > >            Active: failed (Result: exit-code) since Thu
>> > > > > > > 2014-05-01 02:02:37 UTC; 17s ago
>> > > > > > >           Process: 106
>> > > > > > > ExecStart=/usr/libexec/lxc/lxc-startup start (code=exited,
>> > > > > > > status=203/EXEC) Process: 78
>> > > > > > > ExecStartPre=/usr/libexec/lxc/lxc-devsetup (code=exited,
>> > > > > > > status=0/SUCCESS) Main PID: 106 (code=exited,
>> > > > > > > status=203/EXEC)
>> > > > > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>> > > > > >
>> > > > > > That said that lxc-devsetup ran and returned status=0/SUCCESS
>> > > > > >
>> > > > > > >
>> > > > > > >         CGroup: /user.slice/user-0.slice/session-1.scope/system.slice/lxc.service
>> > > > > > >
>> > > > > > >         May 01 02:02:37 nat-1 systemd[1]: Failed to start
>> > > > > > > LXC Container Initialization and Autoboot Code.
>> > > > > > >         May 01 02:02:37 nat-1 systemd[1]: Unit lxc.service
>> > > > > > > entered failed state.
>> > > > > > >         May 01 02:02:37 nat-1 lxc-devsetup[78]: /dev is
>> > > > > > > devtmpfs
>> > > > > > >
>> > > > > > >
>> > > > > > >
>> > > > > > >         On Thu, May 1, 2014 at 8:58 AM, István Király
>> > > > > > > <LaKing at d250.hu> wrote:
>> > > > > > >                 I have no experience with the pre-packed
>> > > > > > > rpm, I usually build it from source ...
>> > > > > > >
>> > > > > > >
>> > > > > > >                 So my questions would be:
>> > > > > > >
>> > > > > > >
>> > > > > > >                 Can you create and start containers -
>> > > > > > > without the service running.
>> > > > > > >                 Does this directory exists, and what if not
>> > > > > > > and you create it? /usr/libexec/lxc/lxc-startup
>> > > > > > >
>> > > > > > >
>> > > > > > >
>> > > > > > >
>> > > > > > >
>> > > > > > >
>> > > > > > >                 On Thu, May 1, 2014 at 5:57 AM, CDR
>> > > > > > >                 <venefax at gmail.com> wrote:
>> > > > > > >
>> > > > > > >                         Dear Friends
>> > > > > > >                         I am using the latest version of LXC
>> > > > > > >
>> > > > > > >                         rpm -qa | grep lxc
>> > > > > > >                         lxc-devel-1.0.3-1.fc20.x86_64
>> > > > > > >                         libvirt-daemon-driver-lxc-1.1.3.4-4.fc20.x86_64
>> > > > > > >                         lxc-libs-1.0.3-1.fc20.x86_64
>> > > > > > >                         lxc-1.0.3-1.fc20.x86_64
>> > > > > > >                         lxc-debuginfo-1.0.3-1.fc20.x86_64
>> > > > > > >
>> > > > > > >                          systemctl start lxc.service
>> > > > > > >
>> > > > > > >
>> > > > > > >                         - Unit lxc.service has begun
>> > > > > > > starting up. Apr 30 15:51:49 nat-1 lxc-devsetup[8036]:
>> > > > > > >                         Creating /dev/.lxc
>> > > > > > >                         Apr 30 15:51:49 nat-1
>> > > > > > > lxc-devsetup[8036]: /dev is devtmpfs
>> > > > > > >                         Apr 30 15:51:49 nat-1
>> > > > > > > lxc-devsetup[8036]: Creating /dev/.lxc/user
>> > > > > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>> > > > > > That line, right there, is an echo out of lxc-devsetup.
>> > > > > > That's obviously there.
>> > > > > >
>> > > > > > >                         Apr 30 15:51:49 nat-1 systemd[8042]:
>> > > > > > > Failed at step EXEC spawning
>> > > > > > >                         /usr/libexec/lxc/lxc-startup: No
>> > > > > > > such file or directory
>> > > > > > >                         -- Subject:
>> > > > > > >                         Process /usr/libexec/lxc/lxc-startup
>> > > > > > > could not be executed
>> > > > > > >                         -- Defined-By: systemd
>> > > > > > >                         -- Support:
>> > > > > > >                         http://lists.freedesktop.org/mailman/listinfo/systemd-devel
>> > > > > > >                         --
>> > > > > > >                         -- The
>> > > > > > > process /usr/libexec/lxc/lxc-startup could not be executed
>> > > > > > > and failed. --
>> > > > > > >                         -- The error number returned while
>> > > > > > > executing this process is 2.
>> > > > > > >                         Apr 30 15:51:49 nat-1 systemd[1]:
>> > > > > > > lxc.service: main process exited,
>> > > > > > >                         code=exited, status=203/EXEC
>> > > > > > >                         Apr 30 15:51:49 nat-1 systemd[1]:
>> > > > > > > Failed to start LXC Container
>> > > > > > >                         Initialization and Autoboot Code.
>> > > > > > >                         -- Subject: Unit lxc.service has
>> > > > > > > failed -- Defined-By: systemd
>> > > > > > >                         -- Support:
>> > > > > > >                         http://lists.freedesktop.org/mailman/listinfo/systemd-devel
>> > > > > > >                         --
>> > > > > > >                         -- Unit lxc.service has failed.
>> > > > > > >
>> > > > > >
>> > > > > >
>> > > > > > Regards,
>> > > > > > Mike
>> > > > > > --
>> > > > > > Michael H. Warfield (AI4NB) | (770) 978-7061 |
>> > > > > > mhw at WittsEnd.com /\/\|=mhw=|\/\/          | (678) 463-0932 |
>> > > > > > http://www.wittsend.com/mhw/ NIC whois: MHW9          | An
>> > > > > > optimist believes we live in the best of all PGP Key:
>> > > > > > 0x674627FF        | possible worlds.  A pessimist is sure of
>> > > > > > it!
>> > > > > >
>> > > > >
>> > > > >
>> > > > >
>> > > > > > _______________________________________________
>> > > > > > lxc-users mailing list
>> > > > > > lxc-users at lists.linuxcontainers.org
>> > > > > > http://lists.linuxcontainers.org/listinfo/lxc-users
>> > > > >
>> > > > >
>> > > > > _______________________________________________
>> > > > > lxc-users mailing list
>> > > > > lxc-users at lists.linuxcontainers.org
>> > > > > http://lists.linuxcontainers.org/listinfo/lxc-users
>> > > >
>> > >
>> > >
>> >
>>
>>
>
> --
> Michael H. Warfield (AI4NB) | (770) 978-7061 |  mhw at WittsEnd.com
>    /\/\|=mhw=|\/\/          | (678) 463-0932 |  http://www.wittsend.com/mhw/
>    NIC whois: MHW9          | An optimist believes we live in the best of all
>  PGP Key: 0x674627FF        | possible worlds.  A pessimist is sure of it!
>
>
> _______________________________________________
> lxc-users mailing list
> lxc-users at lists.linuxcontainers.org
> http://lists.linuxcontainers.org/listinfo/lxc-users


More information about the lxc-users mailing list