我有一个安装了EBS的AWS EC2实例/docker：
root@ip-10-0-0-235:~# lsblk /dev/xvdc1 NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT xvdc1 202:33 0 100G 0 part /docker
然后我/etc/docker/daemon.json与：
{
"graph": "/docker/"
}
更新Docker版本后 - 我必须按照here更改graph至data-root：
{
"data-root": "/docker/"
}
但在那次改变之后 - 我面临另一个错误：
Nov 22 09:15:31 ip-10-0-0-235 dockerd[5671]: time="2017-11-22T09:15:31.168591095Z" level=info msg="libcontainerd: new containerd process, pid: 5679"
Nov 22 09:15:32 ip-10-0-0-235 dockerd[5671]: time="2017-11-22T09:15:32.170033835Z" level=warning msg="failed to rename /docker/tmp for background deletion: %!s(<nil>). Deleting synchronously"
Nov 22 09:15:32 ip-10-0-0-235 dockerd[5671]: time="2017-11-22T09:15:32.172631277Z" level=error msg="[graphdriver] prior storage driver aufs failed: invalid argument"
Nov 22 09:15:32 ip-10-0-0-235 dockerd[5671]: Error starting daemon: error initializing graphdriver: invalid argument
现在安装它的正确方法是什么？
Docker 17.05.0-ce，在Ubuntu 16.04.3上运行。
P.S。在写这个Q期间 - 找到了奇怪的挂载分区：
root@ip-10-0-0-235:~# mount | grep xvd /dev/xvda1 on / type ext4 (rw,relatime,discard,data=ordered) /dev/xvda1 on /docker/aufs type ext4 (rw,relatime,discard,data=ordered) /dev/xvdb1 on /jenkins type ext4 (rw,relatime,data=ordered) /dev/xvdc1 on /docker type ext4 (rw,relatime,data=ordered)
/dev/xvda1 on /docker/aufs和/dev/xvdc1 on /docker。这看起来不正确......