Docker fails to start after install with "loopback attach failed" - docker

Question

I have installed docker-ce from repository following instructions at:
https://docs.docker.com/install/linux/docker-ce/centos/
I receive an error attempting to start docker:
Job for docker.service failed because the control process exited with error code. See "systemctl status docker.service" and "journalctl -xe" for details.
journalctl has the following:
...
dockerd[3647]: time="2018-02-05T14:47:05-08:00" level=info msg="containerd successfully booted in 0.002946s" module=containerd
dockerd[3647]: time="2018-02-05T14:47:05.456552594-08:00" level=error msg="There are no more loopback devices available."
dockerd[3647]: time="2018-02-05T14:47:05.456585240-08:00" level=error msg="[graphdriver] prior storage driver devicemapper failed: loopback attach failed"
dockerd[3647]: Error starting daemon: error initializing graphdriver: loopback attach failed
systemd[1]: docker.service: main process exited, code=exited, status=1/FAILURE
systemd[1]: Failed to start Docker Application Container Engine.
I have seen articles about using something other than loopback devices, but as far as I can tell, those indicate an optimization to be made - and do not imply that the initial startup should fail.
CentOS Linux release 7.4.1708 (Core)

Answer 1

If you run Linux in a VM on Xen, you need to install the kernel and use pygrub (see https://wiki.debian.org/PyGrub) and update to docker version 19.03.0.
install pygrub
1. In your VM execute:
mkdir /boot/grub
apt-get install -y linux-image-amd64
cat > /boot/grub/menu.lst << EOF
default 0
timeout 2
title Debian GNU/Linux
root (hd0,0)
kernel /vmlinuz root=/dev/xvda2 ro
initrd /initrd.img
title Debian GNU/Linux (recovery mode)
root (hd0,0)
kernel /vmlinuz root=/dev/xvda2 ro single
initrd /initrd.img
EOF
2. halt your VM, for example:
xen destroy vm01
3. edit your xen config
for example for your VM /etc/xen/vm01.cfg in your DOM0 (comment out the first two lines and add the last three):
#kernel = '/boot/vmlinuz-4.9.0-9-amd64'
#ramdisk = '/boot/initrd.img-4.9.0-9-amd64'
extra = 'elevator=noop'
bootloader = '/usr/lib/xen-4.8/bin/pygrub'
bootloader_args = [ '--kernel=/vmlinuz', '--ramdisk=/initrd.img', ]
4. start your vm:
xen create /etc/xen/vm01.cfg

Answer 2

I have the same problem in a Debian 9 VM and the same in Debian 8 VM both on the same Debian XEN 4.8 host.
loopback seems not to exist:
# losetup -f
losetup: cannot find an unused loop device: No such device
You can create those with
#!/bin/bash
ensure_loop(){
num="$1"
dev="/dev/loop$num"
if test -b "$dev"; then
echo "$dev is a usable loop device."
return 0
fi
echo "Attempting to create $dev for docker ..."
if ! mknod -m660 $dev b 7 $num; then
echo "Failed to create $dev!" 1>&2
return 3
fi
return 0
}
ensure_loop 0
ensure_loop 0
But this is just a tip to find the right solution, it didn't solve it completely, now since /dev/loop0 exists, I have the error:
Error opening loopback device: open /dev/loop0: no such device or address
[graphdriver] prior storage driver devicemapper failed: loopback attach failed
Update:
I installed apt-get install docker-ce docker-ce-cli containerd.io like described in the latest docs and now with the latest version:
$ docker --version
Docker version 19.03.0, build aeac9490dc
still the same issue:
failed to start daemon: error initializing graphdriver: loopback attach failed
This is the full log:
level=info msg="Starting up"
level=warning msg="failed to rename /var/lib/docker/tmp for background deletion: rename /var/lib/docker/tmp
/var/lib/docker/tmp-old: file exists. Deleting synchronously"
level=info msg="parsed scheme: \"unix\"" module=grpc
level=info msg="scheme \"unix\" not registered, fallback to default scheme" module=grpc
level=info msg="ccResolverWrapper: sending update to cc: {[{unix:///run/containerd/containerd.sock 0 <nil>}
] }" module=grpc
level=info msg="ClientConn switching balancer to \"pick_first\"" module=grpc
level=info msg="pickfirstBalancer: HandleSubConnStateChange: 0xc0005e8660, CONNECTING" module=grpc
level=info msg="pickfirstBalancer: HandleSubConnStateChange: 0xc0005e8660, READY" module=grpc
level=info msg="parsed scheme: \"unix\"" module=grpc
level=info msg="scheme \"unix\" not registered, fallback to default scheme" module=grpc
level=info msg="ccResolverWrapper: sending update to cc: {[{unix:///run/containerd/containerd.sock 0 <nil>}
] }" module=grpc
level=info msg="ClientConn switching balancer to \"pick_first\"" module=grpc
level=info msg="pickfirstBalancer: HandleSubConnStateChange: 0xc0007f5b10, CONNECTING" module=grpc
level=info msg="pickfirstBalancer: HandleSubConnStateChange: 0xc0007f5b10, READY" module=grpc
level=error msg="There are no more loopback devices available."
level=error msg="[graphdriver] prior storage driver devicemapper failed: loopback attach failed"
failed to start daemon: error initializing graphdriver: loopback attach failed
Update 2:
In the end I found out, that pygrub was missing in the VM, which seems to be a new dependency since some version.
This answer was a dead end path, I added another answer, but I leave this here for other users, that have a different problem to get some hints.

Answer 3

I have meet this issue too. I resolved this issue!
In my VMWare workstation, the VM have TWO virtual network interfaces.
I removed one of the virtual network interfaces, and reserved only one.
Start VMWare workstation,start docker service, it works successfully!
I installed docker on CentOS7.6(1810),but when I start docker:
#systemctl start docker
Docker starts failed.
#journalctl -xe
It show some messages like "start daemon: error initializing graphdriver: loopback attach failed".