<p>On Jul 16, 2011 6:26 PM, "Michael H. Warfield" <<a href="mailto:mhw@wittsend.com">mhw@wittsend.com</a>> wrote:<br>
><br>
> On Sat, 2011-07-16 at 23:59 +0300, Ramez Hanna wrote:<br>
> > On Sat, Jul 16, 2011 at 8:27 PM, Michael H. Warfield <<a href="mailto:mhw@wittsend.com">mhw@wittsend.com</a>>wrote:<br>
> ><br>
> > > On Fri, 2011-07-15 at 20:17 +0300, Ramez Hanna wrote:<br>
> > ><br>
> > > << Big Snip >><br>
> > ><br>
> > > > > > thanks a lot for the detailed answer<br>
> > > > > > by the way have you been succesfull in starting a f15 container on<br>
> > > your<br>
> > > > > f15?<br>
> > ><br>
> > > I now have an F15 container working.<br>
> > ><br>
> > > > > > I have been debuggin for 2 hours now<br>
> > > > > > when i start f15 container it screws my host by interfering with my<br>
> > > > > hosts's<br>
> > > > > > systemd which somehow doesn't make sense<br>
> > > > > > and when i use systemd-nspawn i get a bunch of errors and the system<br>
> > > > > doesn't<br>
> > > > > > finish starting<br>
> > > > > > here is a paste of systemd log from systemd-nspawn session<br>
> > > > > > <a href="http://pastie.org/2218625">http://pastie.org/2218625</a><br>
> > > > ><br>
> > > > > I haven't tried it yet. Will see what I can do.<br>
> > > > ><br>
> > > > > Couple of quick questions.<br>
> > > > ><br>
> > > > > 1) You say it screws your host if you don't uses nospawn. What<br>
> > > happens?<br>
> > ><br>
> > > > host console is not useable, random issues around missing characters when<br>
> > > i<br>
> > > > type<br>
> > > > unable to login on other terminals because i cannot type<br>
> > > > and i see so many systemd logs on the console<br>
> > ><br>
> > > I have a very strong suspicion that systemd is not going to be<br>
> > > compatible with running in a container because it wants to set up and<br>
> > > managed cgroups in the container which it can not do.<br>
> > ><br>
> > > When I try to start it with systemd, the first process doesn't even seem<br>
> > > to come up (number of tasks is 0) and then the host can not remove the<br>
> > > container even after I've done an lxc-stop on it. But that's when I'm<br>
> > > logged in and running lxc-start from an ssh terminal Window. If I start<br>
> > > it from a real ttyX console then I get all sorts of startup messages<br>
> > > from the container and the consoles are hosed up like the console in the<br>
> > > container has gotten crosswise with the console in the host. Things try<br>
> > > to initialize but all sorts of things time out and eventually I have to<br>
> > > reset the host with an Magic SysRq sequence.<br>
> > ><br>
> > > Gave up on systemd.<br>
> > ><br>
> > > > > 2) Have you disabled the sys_admin cap by dropping it in that<br>
> > > container?<br>
> > > > > I find that causes me all sorts of grief.<br>
> > > > ><br>
> > > > i will try that<br>
> > ><br>
> > > Don't. It wouldn't do any good and causes lots of other problems (for<br>
> > > me at least).<br>
> > ><br>
> > > > > 3) Was this a fresh template build or did you upgrade an F14 machine to<br>
> > > > > F15 (I was going to use "yum --releasever=15 distro-sync" in one of my<br>
> > > > > running F14 containers).<br>
> > ><br>
> > > > yes fresh install<br>
> > ><br>
> > > Here's what I've done and now gotten an F15 container to work.<br>
> > ><br>
> > > I started out with an F14 container and upgraded it to F15 using the<br>
> > > "yum --releasever=15 distro-sync" method. I was able to reproduce your<br>
> > > problems above and thought there may be some conflicts over cgroups so I<br>
> > > decided to disable systemd.<br>
> > ><br>
> > > If it's not present (it wasn't for me) install upstart into the<br>
> > > container from the host using "yum --installroot={your VM root}<br>
> > > upstart".<br>
> > ><br>
> > > Next cd to {your VM root}/sbin and rm init (which is symlinked<br>
> > > to ../bin/systemd) and symlink it to upstart (which is in sbin).<br>
> > ><br>
> > > This got me almost there. The machine was starting but I was having<br>
> > > your funky console problem and I realized (largely because I'm working<br>
> > > on other related problems) that it was the ptmx device causing this. It<br>
> > > was mapping incorrectly.<br>
> > ><br>
> > > So, cd to {your VM root}/dev and rm ptmx if it's a hard device and not a<br>
> > > symlink. Then symlink pts/ptmx to ptmx. If you started with some sort<br>
> > > of template, this may already be done and you may not run into this<br>
> > > problem at all.<br>
> > ><br>
> > > Now you should be able to fire your F15 container up.<br>
> > ><br>
> > > Also find the lines in /etc/init.d/halt that remount file systems ro or<br>
> > > you'll screw your /dev/pts fs in the host when you shut that container<br>
> > > down or reboot it (and, no, newinstance is not helping with that<br>
> > > problem).<br>
> > ><br>
> > > Regards,<br>
> > > Mike<br>
> > > --<br>
> > > Michael H. Warfield (AI4NB) | (770) 985-6132 | mhw@WittsEnd.com<br>
> > > /\/\|=mhw=|\/\/ | (678) 463-0932 |<br>
> > > <a href="http://www.wittsend.com/mhw/">http://www.wittsend.com/mhw/</a><br>
> > > NIC whois: MHW9 | An optimist believes we live in the best of<br>
> > > all<br>
> > > PGP Key: 0x674627FF | possible worlds. A pessimist is sure of it!<br>
> > ><br>
><br>
> > it is very clear to me that systemd is interfering with the host's systemd<br>
> > your solution of running f15 is not much different than running a f14<br>
> > container (as systemd is the major diff)<br>
> > systemd-nspawn can start systemd inside a "light weight" container<br>
> > i think the problem is related to the fact that when lxc starts teh cgroup<br>
> > is on the root of the tree<br>
> > while it should have been under the user's tree<br>
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^<br>
><br>
> I'm not so sure I understand what you mean by that last line. What<br>
> "user's tree" are you referring to?<br>
><br>
> > maybe serge can say somethiing about this<br>
><br>
> Maybe, maybe not.<br>
><br>
> The cgroup mounts are where systemd is putting them, not where lxc is<br>
> putting them. As it is, lxc is not "starting" the cgroup anywhere, it's<br>
> just using them where they are found. And systemd-nspawn has nothing to<br>
> do with lxc.<br>
><br>
> Seems to me that where you're trying an F15 guest on an F15 host with<br>
> cgroups mounted in the host by systemd where systemd wants them and then<br>
> using systemd-nspawn to fire up the container, you've completely cut lxc<br>
> out of the loop here? What does that have to do with lxc at this point?<br>
><br>
> I just read up on systemd-nspawn (which I had never tried before) and it<br>
> sounds like a really primitive lxc competitor comparable to lxc-start<br>
> only without the configuration stuff and helper utilities.<br>
><br>
> Tried it out myself firing up the same Faces container only under<br>
> systemd-nspawn instead of lxc-start and execing /bin/systemd instead<br>
> of /sbin/init (upstart) and yeah it face-planted with the same errors<br>
> you encountered. It also wouldn't start the same container<br>
> running /sbin/init but I suspect that was more because of missing<br>
> mounted file systems like sys and proc that lxc-start is so nicely<br>
> handling for us.<br>
><br>
> I would say that problems with systemd-nspawn are systemd-nspawn's and<br>
> not ours here. After looking at it and trying it out, I don't think<br>
> I'll try it again. Even the man pages on it mentions "systemd-nspawn -<br>
> Spawn a namespace container for debugging, testing and building".<br>
> Doesn't sound like a production tool to me.</p>
<p>Yeah its not meant to be a container hypervisor ... at the moment at least. I know Lennarts comments on on were of the opinion that over time it could very well grow more features though.</p>
<p>In my experience at least, systemd has been nothing but pure awesome. I've used it extensively and the level of unit customization is unreal ... you could very well use pure units alone to start/stop/react to container events, and construct/start/manage them on fly.</p>
<p>At and rate it *should* would in a cgroup just fine (I haven't tries w/lxc directly tho) and this will improve; for example, when it detects itself running in a cgroup it calls exit() on shutdown vs. hanging like sysvinit. IMO sysvinit is about as dumb as an app can get -- I could do its job with a 15 line bash script ... I think systemd will soon be the tool of choice here.</p>
<p>C Anthony [mobile]<br>
</p>