[CRIU] overmount confusion

[Pavel Emelyanov]

On 04/01/2015 06:02 PM, Christopher Covington wrote:
> Hi Pavel,
> 
> On 04/01/2015 03:16 AM, Pavel Emelyanov wrote:
> 
>> Yes, that's the strategy for restore. As far as the dump is concerned -- we
>> only need to "get" one in terms of reading the information from /proc. One
>> exception from this rule would be tmpfs which we need to tar. For such cases
>> we can use the former technique -- move the overmounted mountpoints aside
>> temporarily.
> 
> Why is tmpfs tarred?

And gzipped :)

Because we consider "dump" to be a snapshot-like operation. After it the
processes AND the environment can either a) be killed or b) continue
running and after restore things should be restored in the state they
used to be. Also for live migration we should restore tmpfs from somewhere
on the destination node.

having said that -- if tmpfs is killed after dump or is modified after dump
or is not transferred to destination node we have no chances to restore
things properly. IOW tmpfs acts as a ghost file, so we have to take it
into images.

-- Pavel