[Lxc-users] regarding lxc "states" available to lxc-monitor or lxc-wait usage

[Vallevand, Mark K]

Actually, I've had good success using Unix domain named sockets for communications between programs in containers and host.  Perhaps they are in a shared name space.  But, don't change it.  :-)  It works.

It's not a generic solution for reporting status.


Regards.
Mark K Vallevand   Mark.Vallevand at Unisys.com
May you live in interesting times, may you come to the attention of important people and may all your wishes come true.
THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers.


-----Original Message-----
From: Serge Hallyn [mailto:serge.hallyn at ubuntu.com] 
Sent: Thursday, May 09, 2013 6:14 PM
To: Vallevand, Mark K
Cc: Guido Jäkel; bmullan.mail at gmail.com; lxc-users
Subject: Re: [Lxc-users] regarding lxc "states" available to lxc-monitor or lxc-wait usage

Quoting Vallevand, Mark K (Mark.Vallevand at UNISYS.com):
> I don't think signals to the container init process is the answer.  You can run a single program in a container.  When that program reaches a certain state, it could report that it is operational.  I don't know what the correct way to do that report using a new lxc feature.
> 
> Currently, my program creates a file that indicates it is operational.  My container start-up code looks for that file before it assumes that the container is ready.

Right, that's what I'd probably do.  We can't really pass the info over
a socket since the container is likely in a separate netns.  Signals
require that (a) the container's init be in on it, and (b) that
container init get the signals and (c) have a way to pass them on.

Solving this in a generic way doesn't seem easy.