Em 05-01-2011 23:40, Sukadev Bhattiprolu escreveu: > | - https://ckpt.wiki.kernel.org/index.php/Link-LXC-USERCR > > Thats a good start. Indeed. But the first paper is AWESOME for the understanding of CR. :-) > | - http://sourceforge.net/mailarchive/message.php?msg_id=25136852 > | > | in my reading list, but this > | > | - http://lxc.sourceforge.net/patches/lxc+usercr/lxc-patches/ > | - http://lxc.sourceforge.net/patches/lxc+usercr/usercr-patches/ > > where did you find these broken links ? If you can let me know, I will > fix the links. But as mentioned in In [lxc-users] archives: http://www.mail-archive.com/lxc-users@lists.sourceforge.net/msg00283.html > https://ckpt.wiki.kernel.org/index.php/Link-LXC-USERCR > > there are sub-directories (eg lxc-0.7.2) for each version of LXC. > > You may also want to refer to this mail: > > http://lists.linux-foundation.org/pipermail/containers/2010-November/025964.html While the PDF made me understand how things work, wiki + this link (and much googling =) got me in the right track, thanks!
> | > | for example, is not available. > | > | I'm sending this message in the hope that my job would be useful and could > | be done in a way that it benefits the community. So, I would like to ask: > | > | 1) Working on that would help something on, somebody in or someway the > | project? > > You can try out the patches, report problems and if possible submit fixes for > any problems you encounter :-) It's my next course of action! > | 3) Is there a "reference model" as how that functionality should be built? > > Not sure what you mean by a reference model. That was a bad question from my part. I was looking for a roadmap or TODO list. But I found https://ckpt.wiki.kernel.org/index.php/Checklist :-) > | 5) There is something like that already built: vzmigrate. > > | Are you OK with > | the concept of copying a frozen machine to it's new place through ssh? > > Not sure exactly what you mean, but yes you can ssh the checkpoint image > to a different machine with similar configuration and restart the container. In OpenVZ, you migrate containers between physical nodes using "vzmigrate". That command freezes the container, saves its state and scp the state + the respective filesystem to another machine, starts it in the new environment and stops it in the former one. Linux-CR seems concerned in doing the first part (saving state), but it doesn't seem to care much about how or if the ckpt will be transferred to another box. I would like to know if you think that a tool that automates the transfer of ckpt + filesystem to another box and subsequent restart (analog to vzmigrate) would be a good newcomer. That said, I'm already looking to distributed FS's and friends (like Gluster) as another solution to have a container migrated in a fast and automatic way between multiple nodes. If containers were stored on '/srv/lxc' (that being a Gluster volume) and we had containers rootfs in '/srv/lxc/CONT/rootfs', then doing something like 'lxc-checkpoint -n CONT -s /srv/lxc/CONT/cont.ckpt' would allow any node in the Gluster volume to restart it. Will try that, what may produce useful documentation. If you have any suggestions, warnings or concerns about that, I would be pleased to know. > | Or is LXC trying to mimetize as little as possible other projects? > > I don't understand what you mean by this. That was a silly, silly question from my part. As there were other container-based virtualization solutions around, I thought that getting inspiration from existing tools and another frameworks (like the ssh transfer in vzmigrate/OpenVZ) would not be seem as a good way to start. But in my readings about LXC I saw many references to other projects and it's developers. That's the true free software beauty =) Thanks for your reply, I will try to report any good progress and share its results. Bill. ------------------------------------------------------------------------------ Learn how Oracle Real Application Clusters (RAC) One Node allows customers to consolidate database storage, standardize their database environment, and, should the need arise, upgrade to a full multi-node Oracle RAC database without downtime or disruption http://p.sf.net/sfu/oracle-sfdevnl _______________________________________________ Lxc-devel mailing list Lxc-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lxc-devel
