Hello everyone,
I’m running LXD 3.0.3 installed via apt on Ubuntu 18.04, using an LVM storage pool. I have an image that is 11GB when unpacked, and I’m trying to launch a container with it. I’m aware that I can modify the default size of a container by setting volume.size, but I prefer to use a profile rather than change the global configuration of my storage pool.
This is the configuration of my storage pool:
$ lxc storage list
+---------+-------------+--------+--------------------------------+---------+
| NAME | DESCRIPTION | DRIVER | SOURCE | USED BY |
+---------+-------------+--------+--------------------------------+---------+
| default | | lvm | /var/lib/lxd/disks/default.img | 14 |
+---------+-------------+--------+--------------------------------+---------+
$ lxc storage show default
config:
lvm.thinpool_name: LXDThinPool
lvm.vg_name: default
size: 100GB
source: /var/lib/lxd/disks/default.img
volume.size: 10GB
description: ""
name: default
driver: lvm
used_by:
[ redacted ]
status: Created
locations:
- none
I created a new profile by copying the default profile:
$ lxc profile copy default myprofile
I tried to use the command listed in this thread on GitHub to set the size of the profile, but lxc did not seem to recognize the command:
$ lxc profile set myprofile root size 20GB
Description:
Set profile configuration keys
Usage:
lxc profile set [<remote>:]<profile> <key> <value> [flags]
Global Flags:
--debug Show all debug messages
--force-local Force using the local unix socket
-h, --help Print help
-v, --verbose Show all information messages
--version Print version number
Error: Invalid number of arguments
But I was able to edit it manually to add the “size: 20GB” line:
$ lxc profile edit myprofile
$ lxc profile show myprofile
config: {}
description: Default LXD profile
devices:
eth0:
name: eth0
nictype: bridged
parent: lxdbr0
type: nic
root:
path: /
pool: default
size: 20GB
type: disk
name: myprofile
When I launch a small image using this profile, it does seem to set the size of the container correctly:
$ lxc launch ubuntu:18.04 test -p myprofile
Creating test
Starting test
$ lxc exec test -- bash
root@test:~# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/default/containers_test 20G 342M 19G 2% /
none 492K 0 492K 0% /dev
udev 2.9G 0 2.9G 0% /dev/tty
tmpfs 100K 0 100K 0% /dev/lxd
tmpfs 100K 0 100K 0% /dev/.lxd-mounts
tmpfs 3.0G 0 3.0G 0% /dev/shm
tmpfs 3.0G 8.1M 2.9G 1% /run
tmpfs 5.0M 0 5.0M 0% /run/lock
tmpfs 3.0G 0 3.0G 0% /sys/fs/cgroup
root@test:~# dd if=/dev/zero of=testfile bs=1M count=11000
11000+0 records in
11000+0 records out
11534336000 bytes (12 GB, 11 GiB) copied, 222.289 s, 51.9 MB/s
root@test:~# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/default/containers_test 20G 12G 7.6G 60% /
none 492K 0 492K 0% /dev
udev 2.9G 0 2.9G 0% /dev/tty
tmpfs 100K 0 100K 0% /dev/lxd
tmpfs 100K 0 100K 0% /dev/.lxd-mounts
tmpfs 3.0G 0 3.0G 0% /dev/shm
tmpfs 3.0G 8.1M 2.9G 1% /run
tmpfs 5.0M 0 5.0M 0% /run/lock
tmpfs 3.0G 0 3.0G 0% /sys/fs/cgroup
However, when I try to unpack my 11GB image into a container, the process fails:
$ lxc launch 11gb-image test2 -p myprofile
Creating test2
Error: Unable to unpack image, run out of disk space (consider increasing your pool's volume.size)
The only possibility I can imagine is that LXD is creating a default-size (10GB) volume, unpacking the image into the volume, and then resizing the volume, which seems counterintuitive (as well as inefficient). Is launching large images using a profile not supported, or is this behavior a bug?