thesandlord / google-cloud-swarm Goto Github PK

It will be good if docker-machine rm -f $PREFIX-manager-$i && under swarm-up.sh does just a quick check and report back that the $PREFIX-manager1 doesn't exist as an information rather than the script throwing huge number of error messages:

Error removing host "collab-manager-1": Host does not exist: "collab-manager-1"
Can't remove "collab-manager-1"
About to remove collab-worker-1
Error removing host "collab-worker-1": Host does not exist: "collab-worker-1"
Can't remove "collab-worker-1"
About to remove collab-worker-2
Error removing host "collab-worker-2": Host does not exist: "collab-worker-2"
Can't remove "collab-worker-2"

Otherwise, a good timesaving tool.

@thesandlord Hey thanks for this projects its really cool! I was just testing it out this afternoon and it seems like the Install Docker step never returns finished it was just returning pending for about 15 mins. Do you have any idea why that might be? Looks to be like the manager and workers were up I ssh into the nodes and did a simple sudo docker ps which worked.

Under swam-up.sh, --listen-addr $masterip:2377 is missing due to which docker swarm doesnt get created. The right command is:

docker swarm init --listen-addr $SWARM_MANAGER_INTERNAL_IP:2377 --secret $SWARM_SECRET

When a cluster is resized down, the shutdown-script that removed the node from the swarm is not being triggered. This leaves nodes in the swarm that are "Status: Down".

Luckily it doesn't seem to break anything in Swarm.

I recommend test.docker.com instead of experimental.docker.com which gets the latest 1.12-rc3 release installed and swarm mode working. While I used experimental.docker.com, it pulled 1.11 instead of 1.12.

I think you should use docker swarm join --manager ... in the loop that creates the additional managers. Without that option, they will join the cluster as workers.