Deploying

Deploying a THORNode and its associated services.

Deploy THORNode services

Now you have a Kubernetes cluster ready to use, you can install the THORNode services.

Requirements

  • Running Kubernetes cluster

  • Kubectl configured, ready and connected to running cluster

Steps

Clone the node-launcher repo. All commands in this section are to be run inside of this repo.

git clone https://gitlab.com/thorchain/devops/node-launcher
cd node-launcher
git checkout master

Install Helm 3

Install Helm 3 if not already available on your current machine:

make helm
make helm-plugins

Tools

To deploy all tools, metrics, logs management, Kubernetes Dashboard, run the command below.

make tools

You need to give the deployment a namespace name, thorchain is used in the example below.

If you are successful, you will see the following message:

If there are any errors, they are typically fixed by running the command again.

make help will list all commands available. See here for more information.

Deploy THORNode

It is important to deploy the tools first before deploying the THORNode services as some services will have metrics configuration that would fail and stop the THORNode deployment.

You have multiple commands available to deploy different configurations of THORNode. The commands deploy the umbrella chart thornode-stack in the background in the Kubernetes namespace thornode by default.

NET=mainnet TYPE=<type> NAME=<namepsace name> make install e.g.
NET=mainnet TYPE=validator NAME=thornode make install

If successful, you will see the following:

You are now ready to join the network:

Debugging

Use the following useful commands to view and debug accordingly. You should see everything running and active. Logs can be retrieved to find errors:

kubectl get pods -n thornode
kubectl get pods --all-namespaces
kubectl logs -f <pod> -n thornode

Kubernetes should automatically restart any service, but you can force a restart by running:

kubectl delete pod <pod> -n thornode

Note, to expedite syncing external chains, it is feasible to continually delete the pod that has the slow-syncing chain daemon (eg, binance-daemon-xxx).

Killing it will automatically restart it with free resources and syncing is notably faster. You can check sync status by viewing logs for the client to find the synced chain tip and comparing it with the real-world blockheight, ("xxx" is your unique ID):

kubectl logs -f deploy/binance-daemon -n thornode

CHART SUMMARY

THORNode full stack / chart

  • thornode: Umbrella chart packaging all services needed to run a fullnode or validator THORNode.

This should be the only chart used to run THORNode stack unless you know what you are doing and want to run each chart separately (not recommended).

THORNode services:

  • thornode: THORNode daemon

  • gateway: THORNode gateway proxy to get a single IP address for multiple deployments

  • bifrost: Bifrost service

  • midgard: Midgard API service

Tools

  • prometheus: Prometheus stack for metrics

  • loki: Loki stack for logs

  • kubernetes-dashboard: Kubernetes dashboard

Last updated