Docker Community Forums

Share and learn in the Docker community.

Migrating direct-lvm setup between hosts

docker

(David Dyball) #1

Hi All,

I’ve recently decided to heed the warnings about running devicemapper in loopback mode, so began investigating direct-lvm. I can get it up and running OK, but I’m wondering if it’s possible to move the devicemapper logical volumes from one machine to another without losing the data?

So far when I spin up an EC2 instance and attach a persistent EBS volume and go through the direct-lvm setup it goes fine, but if I terminate the instance, spin up a new one and attach the same EBS volume docker always wipes the contents of the data and metadata logical volumes every time and I lose any containers and images that I had previously created.

Here’s my user-data script to get it all setup

#!/bin/bash
apt-get update && apt-get install -y lvm2
wget -O - https://get.docker.io | bash
service docker stop
rm -rf /var/lib/docker
echo 'DOCKER_OPTS="--storage-driver=devicemapper --storage-opt dm.datadev=/dev/docker/data --storage-opt dm.metadatadev=/dev/docker/metadata"' > /etc/default/docker
lvscan
lvchange -a y /dev/docker/data
lvchange -a y /dev/docker/metadata
service docker start

Here’s the output that docker engine gives when starting up again:

ERRO[0000] devmapper: Device ID 1 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 2 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 3 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 4 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 5 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 6 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 7 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 8 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 9 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 10 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 11 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 12 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 13 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 14 exists in pool but it is supposed to be unused 
ERRO[0000] devmapper: Device ID 15 exists in pool but it is supposed to be unused 
WARN[0000] devmapper: XFS is not supported in your system. Either the kernel doesnt support it or mkfs.xfs is not in your PATH. Defaulting to ext4 filesystem 
INFO[0000] devmapper: Creating filesystem ext4 on device docker-202:1-149365-base 
INFO[0005] devmapper: Successfully created filesystem ext4 on device docker-202:1-149365-base 
INFO[0005] Graph migration to content-addressability took 0.00 seconds 
INFO[0005] Firewalld running: false                     
INFO[0005] Default bridge (docker0) is assigned with an IP address 172.17.0.0/16. Daemon option --bip can be used to set a preferred IP address 
WARN[0005] Your kernel does not support swap memory limit. 
INFO[0005] Loading containers: start.                   

INFO[0005] Loading containers: done.                    
INFO[0005] Daemon has completed initialization          
INFO[0005] Docker daemon                                 commit=c3959b1 execdriver=native-0.2 graphdriver=devicemapper version=1.10.2
INFO[0005] API listen on /var/run/docker.sock

I can see that it’s finding existing IDs in the volumes, but it just seems to wipe them anyway.

Am I missing something simple, or will devicemapper always re-initialize the LVMs that it’s using when it starts up?

Thanks,
DD