[CRIU] problem migrating containers not on shared storage
Tycho Andersen
tycho.andersen at canonical.com
Thu Sep 4 07:49:28 PDT 2014
On Thu, Sep 04, 2014 at 06:33:30PM +0400, Andrew Vagin wrote:
> On Thu, Sep 04, 2014 at 09:07:50AM -0500, Tycho Andersen wrote:
> > Hi all,
> >
> > I'm trying to migrate containers across two vms using rsync to move
> > their filesystems across. I get the following error:
> >
> > (00.140226) 1: Error (fsnotify.c:126): fsnotify: Can't open file handle for 0x0fd00001:0x0000000000180009: Stale file handle
> >
> > This seems to be because criu is keeping around the inodes of the
> > files that were inotify_add_watch()ed, instead of the path names. Does
> > that sound right? Is there any reason we can't move it to keep the
> > pathnames instead and avoid this problem?
> >
> > Also, it looks like there are some other inode related things in the
> > images. The one in fh.proto looks like it might cause similar
> > problems. Does anyone have any thoughts?
>
> Look at the --force-irmap, it should solve your problem.
It does indeed, thanks for the quick response!
Tycho
> commit edde5fb46163c458798de7bff9485802880f1ea7
> Author: Pavel Emelyanov <xemul at parallels.com>
> Date: Thu Mar 6 14:56:03 2014 +0400
>
> irmap: Add option that forces fsnotify watches paths resolve
>
> When migrating container with copying its FS, the inode numbers
> and thus their handles wil change. This will make the restore of
> inotify/fanotify fail, since they do it via fhandles.
>
> We've already faced the problems with fsnotifies on NFS -- they
> don't work there. To address this an irmap cache is created on
> pre-dump, so to resolve the issue with changed inodes during
> migration, we can force the irmap cache build.
>
> Signed-off-by: Pavel Emelyanov <xemul at parallels.com>
>
>
> >
> > Thanks,
> >
> > Tycho
> > _______________________________________________
> > CRIU mailing list
> > CRIU at openvz.org
> > https://lists.openvz.org/mailman/listinfo/criu
More information about the CRIU
mailing list