remote-rclone:
driver: rclone:latest
driver_opts:
# config: "$(base64 /root/rclone.conf)"
remote: 'GDriveBackupNeoEncrypted:'
allow_other: 'true'
vfs_cache_mode: full
# user-agent: "Docker Ndin"
poll_interval: 0
I can see from the log that the plugin had started up, restored its saved state and successfully finished mounting Google Drive at 18:23:01
, but docker daemon failed to ping its health at 18:23:13
, probably due to unexpected program abort.
Let's see what happens if you start the plugin cleanly.
Stop all compose jobs and docker containers using the remote
Stop the plugin with: docker plugin disable rclone
Remove saved plugin state on the host: sudo rm /var/lib/docker-plugins/rclone/cache/docker-plugin.state
Start the plugin again: docker plugin enable rclone
Check that plugin process is still running as seen from the host: ps -efa | grep "rclone serve docker"
Look in the docker daemon log: sudo cat /var/log/syslog | grep dockerd | less
Is everything OK yet? I hope so...
Start your mount-related containers / compose jobs now.
Check again if the plugin is still running?
If it works, just use it.
If the plugin will die, let's get rid of the docker for a moment and troubleshoot things simplistically:
Install rclone normally on the host as per this link: Install
Copy your plugin'ified rclone config locally: sudo cat /var/lib/docker-plugins/rclone/config/rclone.conf | cat > ./rclone.conf
Run ordinary rclone mount, mimicing the composed options: rclone --config ./rclone.conf mount GDriveBackupNeoEncrypted: /tmp/test-mount-dir --poll-interval=0 --vfs-cache-mode=full --allow-other
Manually check the mount trying to behave like your compose'd program: list directories, copy files, etc
I wish your hunt were fruitful... Let me know below...
It seem the first set of commands is working, I will keep an eye on it, It was fine too for at least a few hours last time, finger crossed.
As you suggested earlier, I might submit a documentation update proposal : I thought the uninstall/reinstall of the plugin would clear the state. I will submit an update with the state removal command as an addition in the troubleshooting section, with of course an accompagniing warning of its destructive side effect.
In any case thanks for the help
FWIW: I had a similar problem which appeared following a server reboot cycle and one of my containers using the rclone plugin failed to auto-start.
Here was the state of the plugin:
# docker plugin ls
ID NAME DESCRIPTION ENABLED
683199f0659d rclone:latest Rclone volume plugin for Docker false
I then attempted to enable:
# docker plugin enable rclone
Error response from daemon: dial unix /run/docker/plugins/683199f0659d741976aae8c5c08c624c7bc2704ace3bfe93e8ab693c95a7e9f3/rclone.sock: connect: no such file or directory
And then went through your steps above and the problem was resolved. Will keep an eye on things and report back if I have more details or repro steps.