Assuming the Docker daemon is restarted automatically by whatever init.d or systemd like process when the OS is restarted, what is the preferred way to restart one or more Docker containers? For example I might have a number of web servers behind a reverse proxy or a database server.
CoreOS uses systemd to manage long running services:
if you start the daemon with docker -d -r
, it will restart all containers that were running prior the daemon stopped.
This will become the default behavior in the next release.
CoreOS uses systemd to manage long running services:
What worked for me is to add --restart='always'
to the container {run -d ...} command
For people want to auto restart a docker container, but didn't specify --restart
flag (default to 'no') while running it, you can use docker update command to add one of the following three other options:
- on-failure
- unless-stopped
- always
See this post for the details. People have problem with always
restart flag on, can consider using either on-failure
or unless-stopped
option.
The only documentation I've seen is Docker's Host Integration docs which are a bit light on details, etc.
Basically, it suggests starting the daemon with -r=false
and using systemd (or upstart if you're using something other than CoreOS).
Used Restart
and RestartSec
to make it work:
# Restart after crash
Restart=on-failure
# Give the service 10 seconds to recover after the previous restart
RestartSec=10s
View the documentation.
© 2022 - 2024 — McMap. All rights reserved.
docker inspect
command seems to imply that the restart attribute is part and parcel to the image once started (akin to ports exposed). As such, once you define an auto-restart attribute, you can't change your mind :( – Hagertydocker update --restart no [CONTAINER]
to change config afterdocker run
– Effrontery