lxczfs: small fixes

- We expect destroy to fail in zfs_clone() so try to silence it so users are
  not irritated when they create zfs snapshots.
- Add -r recursive to zfs_destroy(). This code is only hit when a) the
  container has no snapshots or b) the user calls destroy with snapshots. So
  this should be safe. Without -r snapshots will remain.

Signed-off-by: Christian Brauner <christian.brauner@canonical.com>

src/lxc/bdev.c

@@ -660,6 +660,9 @@ static int zfs_clone(const char *opath, const char *npath, const char *oname,
 		if ((pid = fork()) < 0)
 			return -1;
 		if (!pid) {
+			int dev0 = open("/dev/null", O_WRONLY);
+			if (dev0 >= 0)
+				dup2(dev0, STDERR_FILENO);
 			execlp("zfs", "zfs", "destroy", path1, NULL);
 			exit(1);
 		}
@@ -740,7 +743,7 @@ static int zfs_destroy(struct bdev *orig)
 		return -1;
 	*p = '\0';
 
-	execlp("zfs", "zfs", "destroy", output, NULL);
+	execlp("zfs", "zfs", "destroy", "-r", output, NULL);
 	exit(1);
 }