On Sun, Aug 26, 2012 at 3:06 AM, Dan Kegel <d...@kegel.com> wrote: > Trying the same thing with your centos template: > cd /usr/lib/lxc/templates/ > wget > https://raw.github.com/fajarnugraha/lxc/centos-template/templates/lxc-centos.in > mv lxc-centos.in lxc-centos > lxc-create -t centos -n democ6 > lxc-start -n democ6 > resulted in a session without working network.
What host are you using? An ubuntu host will automatically add a networking section like this, which works lxc.network.type=veth lxc.network.link=lxcbr0 lxc.network.flags=up The default networking config section from lxc-fedora (upstream version), when used in Ubuntu host would add an additional interface in the container (because it's basically a separate, additional networking section) and will result in "failed to rename vethXXXXXX->eth0 : File exists" error, so I commented it out. Which is why I'm surprised if you say lxc-fedora works for you, because its default networking section should be the same (i.e. should also cause "eth0: File exists" error). > According to ifconfig inside the Centos session, eth0 > didn't have an ip address. I dimly recall that the network > doesn't start by default on Centos desktops, so I did > /etc/init.d/network start > inside the centos session. That at least got eth0 an ip address, Weird. I tested it on my ubuntu host, and guest container networking is up automatically. Try replacing your container config networking section (lxc.network.*) with the ones from your working ubuntu container, or the one I pasted earlier (i.e. you should NOT have any line that says "lxc.network.name = eth0", only those three lines above for lxc.network.*). > but dns still wasn't working. And it looks like part of the machinery > is missing: > # ls -l /etc/resolv.conf > lrwxrwxrwx 1 root root 29 Mar 12 12:45 /etc/resolv.conf -> > ../run/resolvconf/resolv.conf > Adding a real file there made dns work, and I was able to do simple > networking. This one is partly my fault. Ones a template is used, the resulting rootfs is cached on /var/lib/lxc, and in the case of centos/fedora, it will be updated (yum update) every time you use it to install new containers with that template. Doing "yum --installroot" for update is bad, since the host's yum version may do "bad stuff" (e.g. uses incompatible rpmdb version), so I changed it to "chroot ... yum". However for that to work, I need a working resolv.conf inside the rootfs, so I simply do a "cp -a" from the host's resolv.conf, which works for me because I uninstalled resolvconf. In your case it resulted in a non-working resolv.conf :) Since you already have a working /etc/resolv.conf (i.e. NOT a symlink) inside the container, can you try rebooting the container? If it works (i.e. got networking on boot), I can push a simple fix (which basically would just change "cp -a" to just "cp"). -- Fajar ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Lxc-users mailing list Lxc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lxc-users
