I have database container in swarm network. It has its own node and it is not changing.
The unexpected behavior has started when I add or remove some network for my db container:
For example, I added traefik network:
my_db:
volumes:
db1:/var/lib/postgresql
networks:
- my_project
- traefik # added
Then I run docker swarm deploy
and my database is empty :(
The same behavior is when I remove this network. Both networks are overlay. I tested also on another swarm (which has only one node) and another project - the behavior is the same - database erased.
I checked volumes one node with the only db and portainer which is deployed globally:
docker ps
CONTAINER ID IMAGE
16a78f8b609d postgres:11.3
b71166ae4601 portainer/agent:latest
docker volume ls
DRIVER VOLUME NAME
local 4c3bd578714937e3a5f97e9c5a659a36ad7b9fd0921f57a303f33189275bc997
local 0252ae24cb58f0bd3fec93660f2d09d339d7d579b6d0f9fee5b49e7def9b288b
local 413363adb68602ef02928462e33c53d083a1c507af32c6312632bac68b23cd0a
local c6b042815270a5d37aa710923f69aae9e4ace7c0a0e3a402d0ed3f267acc1923
local cf803014263516e023a208537658c4e7d79d61b5e42fb84d696dfa64fda65c78
local my-project_db1
It seems that older data is in first volumes with hash volume name, and my-project_db1
is recreated when network is added/removed.
I also tried using name in volumes
section - the same.
Problem has been solved by using answer from: How to persist data in a dockerized postgres database using volumes
Adding data in db1:/var/lib/postgresql/data
solved this behavior