Skip to content

Latest commit

 

History

History
101 lines (70 loc) · 3.5 KB

README.md

File metadata and controls

101 lines (70 loc) · 3.5 KB

Cluster Setup on Compute Engine

Steps to set up a Docker Swarm cluster with Google Compute Engine

Note: This is not the best way to create a Swarm cluster on GCP. When the setup script supports Managed Instances Groups, clusters will be more robust and easier to scale.

Prerequisites

Step 1: Configure Setup Script

git clone https://github.com/thesandlord/google-cloud-swarm.git

Specify 3 workers in options.sh

sed -i bak -e s/NUM_WORKERS=2/NUM_WORKERS=3/g options.sh

Step 2: Create Cluster

./swarm-up

Step 3: Add Extra Node to the Cluster

Start a new shell and load environment variables:

sh

source options.sh

Create Node:

docker-machine create $PREFIX-worker-4 \
        --engine-install-url $ENGINE_INSTALL_URL \
        -d google \
        --google-machine-type $WORKER_MACHINE_TYPE \
        --google-zone $WORKER_ZONE \
        --google-disk-size $WORKER_DISK \
        --google-tags $TAGS \
        --google-project $PROJECT_ID

Join the Swarm:

SWARM_MANAGER_INTERNAL_IP=$(gcloud compute instances describe $PREFIX-manager-1 \
    --zone $MANAGER_ZONE --format=text | \
    grep '^networkInterfaces\[[0-9]\+\]\.networkIP:' | \
    sed 's/^.* //g')

eval $(docker-machine env $PREFIX-worker-4) && \
    docker swarm join $SWARM_MANAGER_INTERNAL_IP:2377 --secret $SWARM_SECRET

Exit the shell:

exit

Step 4: Remove Extra Node from the Cluster

$(source options.sh && docker-machine rm -f $PREFIX-worker-4)

$(source options.sh && docker node rm $PREFIX-worker-4)

##Optional - Drain node before deleting: This shouldn't be strictly necessary, as swarm will detect a node is deleted and automatically reschedule work.

Step 4: Get list of Nodes

docker nodes ls

ID NAME MEMBERSHIP STATUS AVAILABILITY MANAGER STATUS
7s21m2bcz5i1ks8dv9hcmfwvk my-swarm-worker-1 Accepted Ready Active
9xd6pzp2fv5is3a69ng0ms7s5 my-swarm-worker-2 Accepted Ready Active
by2uchvhbzx0ybgj11s5y7cto * my-swarm-manager-1 Accepted Ready Active Leader
czaeu4mdifnvkv3ld1mh90xek my-swarm-worker-4 Accepted Ready Active

Step 5: Drain a node

I just picked a node at random

docker node update --availability drain my-swarm-worker-4

docker nodes ls

ID NAME MEMBERSHIP STATUS AVAILABILITY MANAGER STATUS
7s21m2bcz5i1ks8dv9hcmfwvk my-swarm-worker-1 Accepted Ready Active
9xd6pzp2fv5is3a69ng0ms7s5 my-swarm-worker-2 Accepted Ready Active
by2uchvhbzx0ybgj11s5y7cto * my-swarm-manager-1 Accepted Ready Active Leader
czaeu4mdifnvkv3ld1mh90xek my-swarm-worker-4 Accepted Ready Drain

Step 6: Delete Node and Resize Cluster

docker-machine rm -f my-swarm-worker-4

docker node rm my-swarm-worker-4