@substrate/node-watcher

@substrate/node-watcher

Long living script that reads from an archive node and loads select information to a Postgresql DB, a.k.a ETL.

The tasks specific to Nomidot are in src/tasks

The script itself lives in nodeWatcher.ts, and anyone can write their own tasks suited to their needs as long as they conform to the Task interface. It is agnostic of what server or db one decides to use.

Instructions to run locally:

1. Push up the Postgresql and Prisma images: docker-compose up -d
2. Run the script which populates the database: yarn start

On a Local Kubernetes Cluster (minikube) (writes to CloudSQL)

1. Start local cluster minikube start
2. Make sure kubectl is connected to is kubectl cluster-info
3. Follow same steps as below

On Hosted Google Kubernetes Engine

1. Mount secrets to nodewatcher pod's volume: kubectl create -f path/to/secrets.yaml

1. Create namespaces: kubectl create -f path/to/namespace.yaml

• Make sure you're in the appropriate namespace: kubectl config set-context --current --namespace<namespace>
• Verify: kubectl config view --minify | grep namespace:

1. Create nodewatcher deployment: kubectl create -f path/to/nodewatcher_deployment.yaml
2. Create service: kubectl create -f path/to/service.yaml
3. Get the nodewatcher deployment IP: kubectl describe deployments nodewatcher
4. Set PRISMA_ENDPOINT in nomidotwatcher-deployment.yaml to External IP from step 6.
5. Allow traffic: gcloud compute firewall-rules create allow-nodewatcher-nodeport --allow=tcp:31000

Make a rolling update:

• Docker Tag the new image(s) docker tag node_watcher_main eu.gcr.io/kusama-etl/node_watcher:v1.0
• Push it to container registry docker push eu.gcr.io/kusama-etl/node_watcher:v1.0
• Get the current image name kubectl describe pods ${POD_NAME}
• Set new deployment image kubectl set image ${current_image} ${new_image}
• Confirm with kubectl rollout status deployments/${new_image}

PgAdmin into Cloudsql using Proxy

Cloudsql Proxy

• Make sure Cloudsql Proxy is installed: curl -o cloud_sql_proxy https://dl.google.com/cloudsql/cloud_sql_proxy.darwin.amd64
• Make it executable: chmod +x cloud_sql_proxy
• Locate the credentials.json file containing credentials to the service account.
• Start the proxy: ./cloud_sql_proxy -instances=test-installations-222013:europe-west1:kusama-etl=tcp:0.0.0.0:5432 -credential_file=./key.json
• It should say something like:

  Listening on 127.0.0.1:5432 for test-installations-222013:europe-west1:kusama-etl
  2020/01/30 17:05:37 Ready for new connections

PgAdmin

• Get pgadmin: https://hub.docker.com/r/dpage/pgadmin4/

• Run it: docker run -p 80:80 -e PGADMIN_DEFAULT_EMAIL="[e-mail]" -e PGADMIN_DEFAULT_PASSWORD="****" -d dpage/pgadmin4

• Navigate to localhost:80 in a browser

• Login with the above credentials

• Click "New Server"

• In "Connection" tab, fill in the DB_NAME (e.g. nomidot-node-watcher), DB_USER (e.g. admin), and DB_PASS, as appropriate.

• Make sure SSL mode is set to Disabled. n.b. "Even though the sslmode parameter is set to disabled, the proxy does provide an encrypted connection."

• Click Connect.

Port Forwarding

It is also possible to forward our localhost:4467 to our cluster port :4466 with, port forwarding. This will allow us to go to open the GraphQl interface from our local machine.

1. kubectl get pods --namespace nomidotwatcher-staging
2. kubectl port-forward --namespace nomidotwatcher-staging <the-pod-name> 4467:4466

Then just go to localhost:4467 in a browser and that will forward your requests to the cluster's 4466 port (where the deployed Prisma server lives).

You can also deploy a new prisma schema by setting the PRISMA_ENDPOINT environment variable to be localhost:4467, and running prisma deploy, e.g. PRISMA_ENDPOINT="http://127.0.0.1:4467" START_FROM="66647" yarn prisma deploy

Updating Kubernetes Job

1. Make sure you're logged into gcloud with an account that has access to Container Registry. You can also push to Docker Hub but for the sake of having everything in one place, we'll keep it all in GCR.
2. Run gcloud auth configure-docker
3. Check the last build image version with gcloud container images list-tags eu.gcr.io/test-installations-222013/nomidot_watcher
4. Making sure you're on the appropriate branch, build a new image. docker build -t nomidot_watcher:v1.2.15 .
5. Tag that image in the appropraite format for GCR e.g docker tag nomidot_watcher:v1.2.15 eu.gcr.io/test-installations-222013/nomidot_watcher:v1.2.15
6. Push that image with docker push eu.gcr.io/test-installations-222013/nomidot_watcher:v1.2.15
7. Copy the SHA256 hash
8. Paste it into the job config
9. Create the job kubectl create -f /path/to/config.yaml
10. Check it's running kubectl get jobs, kubectl logs -l job-name={jobname} -n nodewatcher-staging
11. Eventually delete the outdated job too kubectl delete jobs old-job-name