I'm sorry, 16 pm today: about 05, due to the sudden instability docker swarm cluster caused new blog background (currently in the gray release phase) does not work properly, this may have caused trouble, please: 55 to 17 understanding.
When a failure period, two containers new blog background can not start properly.
AME NODE DESIRED STATE CURRENT STATE i_web.1 prod-swarm-w3 Running Assigned 5 minutes ago i_web.2 prod-swarm-w4 Running Assigned 2 hours ago
Having identified the problem, we had to delete and re-stack deployment operations.
docker stack rm i ./deploy-production.sh 2.0.6
NAME NODE DESIRED STATE CURRENT STATE i_web.1 prod-swarm-w3 Running Assigned 42 seconds ago i_web.2 prod-swarm-w7 Running Starting 42 seconds ago
After the re-deployment of vessels found on the prod-swarm-w7 node can start, and container issues on the prod-swarm-w3 node remains the same, thereby confirming a prod-swarm-w3 node out of the question, then immediately uninstall the node.
docker node update --availability drain prod-swarm-w3
After unloading, the new blog back quickly returned to normal.
We have decided to use k8s replace docker swarm, but not k8s cluster deployed at this moment about to say with docker swarm 88, has been a docker swarm pit, then we blame covet save trouble, the election of the container (docker container) was on the wrong boat (docker swarm), we will learn the lessons deeply on the wrong ship.