[PATCH 00/80] Kernel based checkpoint/restart [v18]
daniel.lezcano at free.fr
Wed Sep 30 15:08:28 PDT 2009
Dan Smith wrote:
> DL> Yep, I agree. But you didn't answer the question, what are the
> DL> network resources you plan to checkpoint / restart ? eg. you let
> DL> the container to setup your network, will you restore netdev
> DL> statistics ? the mac address ? ipv4 ? ipv6 ?
> Yes, Yes, Yes, and Yes. I'm making the assumption that the common
> case will be with a veth device in the container and that all of the
> aforementioned attributes should be copied over. In the future case
> where we could potentially have a real device in the container, it
> probably doesn't make sense to copy the mac address.
Be careful with the assumptions ;)
> DL> Is it possible to do a detailed list of network resources you plan
> DL> to CR with the different items you will address from userspace and
> DL> kernel space ?
> I'm sure it's possible, but no, I haven't planned out everything for
> the next year. If you have strong feelings about what should be done
> in user and kernel space, feel free to share :)
I just want to understand what is your plan. If you say "yes I will
checkpoint / restart" ipv4, ipv6, netdev statistics, etc ... you should
be able to give at least a small list of network resources you will
checkpoint and how you will restart them, no ?
> DL> Argh ! I was hoping there was something else than the source code
> The header file makes it pretty clear what is going on,
Certainly for you.
We are a little far away of the sys_checkpoint / sys_restart simple
syscalls we talked about at the cr-minisummit in 2008.
More information about the Containers