<div dir="ltr">A couple of points:<div><br></div><div>1. This patch fixes the restore failure that Mark reported with a simple /sbin/init workload. I haven't taken a deeper look to see what this patch can break.</div><div><br></div><div>2. Regarding fixing the issue in libcontainer, as far as I understand, the problem actually happens at checkpoint time when the pathname is erroneously recorded as <span style="font-size:12.8000001907349px">/null//deleted in mountpoint-11.img. As Andrew mentioned, this happens because /sbin/init deletes /dev/null and recreates it.</span></div><div><br></div><div><span style="font-size:12.8000001907349px">--Saied</span></div><div><span style="font-size:12.8000001907349px"><br></span></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Tue, Aug 11, 2015 at 7:44 AM, Cyrill Gorcunov <span dir="ltr"><<a href="mailto:gorcunov@gmail.com" target="_blank">gorcunov@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On Tue, Aug 11, 2015 at 11:37:59AM +0300, Andrew Vagin wrote:<br>
> On Mon, Aug 10, 2015 at 11:34:27AM +0300, Cyrill Gorcunov wrote:<br>
> > This address only a part of a problem with deleted mountpoint,<br>
> > so rather a fast fix to not block development, but warning<br>
> > issued to nag us to not forget resolve it before the release.<br>
><br>
> The result will be incorrect. Do we really want it? This issue can be<br>
> easy fixed in libcontainer which can mount /dev/null from a host file<br>
> system.<br>
<br>
</span>As I said it fixes a part of a problem and issues a warning.<br>
If docker can handle this without a patch, then sure, no need<br>
for it. But as far as I understand Saied said it helps him.<br>
<br>
(sorry for delay in response, drowned in mails)<br>
</blockquote></div><br></div>