Today I got a docker daemon not responding to commands (docker ps, docker run...)
Even after restarting service
In /var/log/docker I got this:
time="2017-03-01T08:28:58.43946917Z" level=fatal msg="open /var/run/docker/libcontainerd/containerd/81623262351dfc42c5e87aa8df11592a57f2d14a468476620c7c4d6c89de1958/state.json: no such file or directory"
The solution was to stop docker service, remove this /var/run/docker directory and restart docker service
docker stop command send SIGTERM to pid1 inside the container to let a chance to stop gracefully (https://www.ctl.io/developers/blog/post/gracefully-stopping-docker-containers/)
Unfortunatly some software ignore SIGTERM and need to be stopped by their own command (hello opendj)
To do so, in the Dockerfile, in the CMD, I use a script like this :
/thecommand/tostart/thesoftware/inbackground
trap "/thecommand/tostop/thesoftware" SIGTERM
while true; do sleep 1; done
so my script is run with PID1, docker stop sends SIGTERM, my script catch (trap) the signal, and run the command to stop gracefully
TIL what is this offical repository :
Only debian with some useful package to build dependencies (needed when you install some gem for example)
For example ruby:2.3 Dockerfile use a buildpack-deps image
I don't know if this shit is expected (normal behaviour or not) but this is annoying
bash :
$ toto=tata
$ echo $toto
tata
$ toto='tata'
$ echo $toto
tata
$ toto="tata"
$ echo $toto
tata
avec docker run -e
$ docker run -e toto=tata -it debian:jessie bash
root@5de64152ed05:/# echo $toto
tata
$ docker run -e toto='tata' -it debian:jessie bash
root@99a6b62f595c:/# echo $toto
tata
$ docker run -e toto="tata" -it debian:jessie bash
root@ab7581c30734:/# echo $toto
tata
avec docker run --env-file:
$ cat env.file
toto="tata"
$ docker run -it --env-file env.file debian:jessie bash
root@805480de4a52:/# echo $toto
"tata"
$ cat env.file
toto='tata'
$ docker run -it --env-file env.file debian:jessie bash
root@004b88cc448d:/# echo $toto
'tata'
$ cat env.file
toto=tata
$ docker run -it --env-file env.file debian:jessie bash
root@2ca9908d1cdf:/# echo $toto
tata
avec docker compose:
$ cat docker-compose.yml
version: '2'
services:
foo:
image: debian:jessie
container_name: foo
environment:
- toto1=tata
- toto2="tata"
- toto3='tata'
command: tail -f /var/log/dmesg
docker-compose up -d
docker exec -it foo bash
root@71a9a886cc24:/# echo $toto1
tata
root@71a9a886cc24:/# echo $toto2
"tata"
root@71a9a886cc24:/# echo $toto3
'tata'
Sympa les petites nouveautés!
docker system df
docker system prune
CLI restructured <3
Improved CLI backwards compatibility <3
Compose file version 2.1 and up
Healthcheck configuration can now be done in the service definition using
the healthcheck parameter
Containers dependencies can now be set up to wait on positive healthchecks
when declared using depends_on. See the documentation for the updated
syntax.
Comment perdre 30 min
:set backupcopy=yes
When you want to modify a file, you have two options, each with its benefits and drawbacks.
You can overwrite the file in place. This does not use any extra space, and conserves the hard links, permissions and any other attribute beyond the content of the existing file. The major drawback of doing this is that if anything happens while the file is being written (the application crashes, or the power goes out), you end up with a partially written file.
You can write the new version of the file to a new file with a different name, then move it into place. This uses more space and breaks hard links, and if you have write permissions on a file but not on the directory it contains, you can't do it at all. On the flip side, the old version of the file is atomically replaced by the new version, so at every point in time the file name points to a valid, complete version of the file.
http://vimdoc.sourceforge.net/htmldoc/options.html#%27backup%27
docker run --rm -it -u $(id -u) -v "${PWD}:/data" -w /data ruby:2.3 \
bundle package --all
Note: Environment persistence can cause unexpected side effects. For example, setting ENV DEBIAN_FRONTEND noninteractive may confuse apt-get users on a Debian-based image. To set a value for a single command, use RUN = .
The environment variables set using ENV will persist when a container is run from the resulting image. You can view the values using docker inspect, and change them using docker run --env =.
La deuxieme approche me semble mieux.. Deux étapes pour build seulement :
git clone
docker build
Au lieu de :
git clone
mvn clean install
docker build
So I just need to split my simple CRUD app into 12 microservices, each with their own APIs which call each others’ APIs but handle failure resiliently, put them into Docker containers, launch a fleet of 8 machines which are Docker hosts running CoreOS, “orchestrate” them using a small Kubernetes cluster running etcd, figure out the “open questions” of networking and storage, and then I continuously deliver multiple redundant copies of each microservice to my fleet. Is that it?
-Yes! Isn’t it glorious?
I’m going back to Heroku.
Flamewar started
[Summary, full note in Shaarli]
The beginner's guide to Infrastructure Automation (Medium, @rdsubhas, via cron weekly). "Encore un bon article qui explique où se situe Docker vis à vis de la CM."
Cool en plus c'est open source (docker)
Génial ce qu'on peut faire avec Docker out of the box maintenant :-)
Un lb en Go qui peut "surveiller" les backends comme Docker par exemple pour update sa conf
On the isolated_nw which was user defined, the Docker network feature updated the /etc/hosts with the proper name resolution. Inside of container2 it is possible to ping container3 by name.
La doc évolue vite au fur et à mesure des versions.. en la relisant j'ai appris plein de nouvelles choses
Cette solution semble etre un bon compromis quand on veut du lien bidirectionnel (ce qui n'est pas possible avec les links) :
http://sgillis.github.io/posts/2015-03-23-docker-dns.html
https://github.com/tonistiigi/dnsdock