This bug was fixed in the package lxd - 2.0.1-0ubuntu1~16.04.1

---------------
lxd (2.0.1-0ubuntu1~16.04.1) xenial; urgency=medium

  * New upstream bugfix release (2.0.1) (LP: #1582891)
    - Don't fail to start when shmounts can't be mounted, instead fail
      container startup
    - Invalidate the simplestreams cache on proxy change
    - Write the container's config file on start to the log path directly
    - Fix crash in list due to empty responses (Issue #1903)
    - Fail when removing non-existent profiles (Issue #1886)
    - Document --alias to image import (Issue #1900)
    - Fix "lxc start" and "lxc stop" options (stateful/stateless)
    - Give better error on invalid source stream (simplestreams)
    - Add basic REST API usage example to README.md
    - Fix typo in lxc stop --help
    - Convert lxc-to-lxd to stable supported pylxd API (Issue #1901)
    - Properly log image update failures
    - Better validate and rollback bad images (Issue #1913)
    - Send operation return value through SmartError
    - Fix basic filtering in lxc list (Issue #1917)
    - Tell the user how to launch a container on first start (Issue #1931)
    - Redirect "remote" to "remote:" when not conflicting (Issue #1931)
    - Don't load the LXC config for snapshots (Issue #1935)
    - list: Allow filtering by unset key (Issue #1917)
    - Fix example in lxc launch
    - Update Japanese translation and other po files
    - Fall back to cpuset.cpus on older kernels (Issue #1929)
    - Properly validate the server configuration keys (Issue #1939)
    - Fix daemonConfig handling of storage
    - Don't remove config file on forkmigrate
    - Fix config handling following config validation change
    - Fixed Markdown syntax in documentation
    - Don't fail early when removing disks (Issue #1964)
    - Don't recursively delete devices
    - Don't fail when some unix devices fail to be deleted
    - Use the same config checks for unix-char and unix-block
    - Allow removing when fs object no longer exists (Issue #1967)
    - Do proper logfile expiry (Issue #1966)
    - Make logging a bit more consistent
    - Don't ignore zfs errors
    - Properly update the mode, uid and gid on existing files (Issue #1975)
    - Detect invalid certificate files (Issue #1977)
    - Fix broken apparmor status check
    - Allow on/off as boolean strings
    - Properly validate the container configuration keys (Issue #1940)
    - Don't mask rsync transfer errors
    - Move execPath to a global variable
    - Use custom netcat instead of nc -U for rsync over websocket (Issue #1944)
    - Fix wrong state dir path in migration
    - Don't fail deleting images when the storage delete fails
    - Improve messages in the Japanese translation
    - Add more checks for the criu binary
    - Rework (live) migration tests
    - Make it explicit in documentation that devices on create are optional
    - Properly record the source of all image copies (Issue #2010)
    - Don't mark containers as ERROR while being created (Issue #1988)
    - Cleanup events sent for operations (Issue #1992)
    - Fix ZFS refcounting issues (Issue #1916 and Issue #2013)
    - Propagate snapshot config when copying a snapshot (Issue #2017)
    - Implement lxc config show for snapshots
    - Add Unix socket example to REST API usage

 -- Stéphane Graber <email address hidden>  Tue, 17 May 2016 18:07:37 -0400