Add a hub to an AWS
The idea behind this guide is showcase the process of setting up a kops-based AWS cluster and manually deploy a new hub on top of it using our deployer tool. This is a preliminary but fully functional and semi-automatic process. Eventually, we should be able to fully automate the first hub deployment process as we currently do with the GKE-based hubs.
We will continue working toward a definitive one once we figured out some of the discussions outlined in #431.
Create an AWS kops-based cluster (optional)¶
Follow the instructions in Add a new AWS cluster if you are not deploying against an existing cluster.
Deploy the new AWS hub¶
Follow the steps outlined in To add a new hub with the following modifications:
Generate a new config file for your hubs if there is not an existing one.
You can use one of the existing hub config files as a “template” for your hubs (for example, here is the Farallon Institute config file). You may need to tweak names,
serverIPand singleuser’s images references. Make sure you set up the
profileListsection to be compatible with your kops cluster (ie. match the
node_selectorwith the proper
false. This creates the hubs without trying to give them HTTPS, so we can appropriately create DNS entries for them.
Create a Pull Request with the new entries, and get a team member to review it.
Once you merge the pull request, the GitHub Workflow will detect that a new entry has been added to the configuration file. It will then deploy a new JupyterHub with the configuration you’ve specified onto the corresponding AWS cluster.
Monitor the action to make sure that it completes.
Get the AWS external IP for your hub with (supposing your hub is
kubectl -n staging get svc proxy-public
To perform the above command successfully, you will need to get the kubernetes context. If you are working with a EKS cluster, you can get the kubeconfig with (modulo you get the credential properly configured):
aws eks update-kubeconfig --name=<NAME_OF_THE_CLUSTER> --region=<REGION>
if you are working with a kops cluster, you can get it with:
kops export kubecfg --admin --name <NAME_OF_THE_CLUSTER>.k8s.local --state s3://2i2c-<NAME_OF_THE_CLUSTER>-kops-state
Create a CNAME record for
staging.foo.2i2c.cloudand point it to the AWS external IP.
Wait for about 10 minutes to make sure the DNS records actually resolves properly. If you are deploying
prodhub as well, you will need to repeat this step for
truein the cluster config file so we can get HTTPS.
Repeat steps 4 and 5.
You need to perform the CNAME record update (step 2 - 6) just once, you will not need to perform those steps for further deployment on pre-existing hubs.