Docker for Azure nodes out of space

Expected behavior

expected to be able to resize the disk of nodes
or at least have Docker use the local SSD of the VMs instead of default 29.4GB disk space.

Actual behavior

all nodes are in the form of VMSS, which use Azure Storage Account.
There’s not an Azure “disk” mounted with the nodes (ie. VMs), hence unable to be re-sized.

Additional Information

Since I launched with Azure VM type D5_V2 which has 800 GB local SSD.
when I run df -h , I can see the 800 GB is /dev/sdb1 and mounted on /mnt/resource.
question is… that is not what Docker is using…
docker is using /dev/sda1.
how do I make all nodes use their actual local SSD? (the /dev/sdb1)

swarm-worker000004:~$ df -h
Filesystem                Size      Used Available Use% Mounted on
overlay                  29.4G      1.4G     26.5G   5% /
tmpfs                    27.5G      4.0K     27.5G   0% /dev
tmpfs                    27.5G         0     27.5G   0% /sys/fs/cgroup
/dev/sda1                29.4G      1.4G     26.5G   5% /home
tmpfs                    27.5G    158.3M     27.3G   1% /mnt
tmpfs                    27.5G    158.3M     27.3G   1% /etc
shm                      27.5G         0     27.5G   0% /dev/shm
tmpfs                    27.5G    158.3M     27.3G   1% /lib/firmware
tmpfs                    27.5G    158.3M     27.3G   1% /lib/modules
/dev/sda1                29.4G      1.4G     26.5G   5% /var/log
/dev/sda1                29.4G      1.4G     26.5G   5% /etc/ssh
/dev/sda1                29.4G      1.4G     26.5G   5% /etc/hosts
/dev/sda1                29.4G      1.4G     26.5G   5% /etc/hostname
/dev/sda1                29.4G      1.4G     26.5G   5% /etc/resolv.conf
tmpfs                     5.5G    324.0K      5.5G   0% /var/run/docker.sock
/dev/sda1                29.4G      1.4G     26.5G   5% /var/lib/waagent
tmpfs                    27.5G    158.3M     27.3G   1% /usr/local/bin/docker
/dev/sdb1               800.0G    191.7M    799.8G   0% /mnt/resource

Steps to reproduce the behavior

  1. set up Docker Swarm with the official docker template (edge version)
    https://docs.docker.com/docker-for-azure/#prerequisites

  2. ssh into any node and run df -h

  3. finds docker not using the local SSD of VM, and the disk each node is using is only 29.4GB ( /dev/sda1 ) which runs out immediately after pulling a few images.

@yclliu thanks for reporting this. This is captured in https://github.com/docker/for-azure/issues/19 . We are working on addressing this in an upcoming release.

2 Likes

@ddebroy Thank you very much for the reference link, do you happen to know when this will be fixed?