Running an Algorand node in Fly.io

Step by step guide

We need a directory with the following structure and contents:

algo_node
   + fly.toml
   + run.sh

The file fly.toml looks the following way

app = 'algorand-node'
primary_region = 'cdg'

[experimental]
  cmd = ['/bin/bash', '/run.sh']

[build]
  image = 'algorand/algod:latest'

[[mounts]]
  source = 'node_vol'
  destination = '/algod/data'
  auto_extend_size_threshold = 80
  auto_extend_size_increment = '1GB'
  auto_extend_size_limit = '500GB'

[http_service]
  internal_port = 8080
  force_https = true
  auto_stop_machines = 'off'
  auto_start_machines = true
  min_machines_running = 1
  processes = ['app']

[[files]]
  guest_path = '/run.sh'
  local_path = 'run.sh'

[[vm]]
  size = 'performance-8x'
  memory = '16gb'
  cpu_kind = 'performance'
  cpus = 8

The above configuration relies on the run.sh script, which has this content

cp /node/run/genesis/mainnet/genesis.json /algod/data
/node/bin/algod -l 127.0.0.1:8080

Now is time to launch our application with the command

fly launch

Now you can answer Yes to the following question:

An existing fly.toml file was found for app algorand-node
? Would you like to copy its configuration to the new app?

and No to the following

? Do you want to tweak these settings before proceeding? (y/N) 

You should see an output like:

✓ Configuration is valid
==> Building image
Searching for image 'algorand/algod:latest' remotely...
image found: img_8y6w4zd58kx7v7rn

Watch your deployment at https://fly.io/apps/algorand-node/monitoring

Provisioning ips for algorand-node
  Dedicated ipv6: 2a09:8280:1::4a:fa1b:0
  Shared ipv4: 66.241.124.189
  Add a dedicated ipv4 with: fly ips allocate-v4

Creating a 1 GB volume named 'node_vol' for process group 'app'. Use 'fly vol extend' to increase its size

To check if everything went well open a shell to the deployment with

fly ssh console

Once there execute

/node/bin/goal node status -w 1000

This will show you an output like

Last committed block: 25644
Time since last block: 0.0s
Sync Time: 126.7s
Last consensus protocol: https://github.com/algorandfoundation/specs/tree/5615adc36bad610c7f165fa2967f4ecfa75125f0
Next consensus protocol: https://github.com/algorandfoundation/specs/tree/5615adc36bad610c7f165fa2967f4ecfa75125f0
Round for next consensus protocol: 25645
Next consensus protocol supported: true
Last Catchpoint: 
Genesis ID: mainnet-v1.0
Genesis hash: wGHE2Pwdvd7S12BL5FaOP20EGYesN73ktiC1qzkkit8=

This shows the node is up and synchronizing with the network

If you don't want to continue running the node, execute

fly app destroy algorand-node

The node synchronization progress can be tracked by comparing Last committed block: 25644 with the last block at Pera Explorer. Calculating 25644 / 43466969 we can see the synchronization progress so far is 0.05%.

Explanation

• The run.sh script is needed because we need the genesis.json file in /algod/data. Once the node is running it will create there other files important for its operation, including blockchain data.

• Since /algod/data contains blockchain data we need a volume, i.e. permanent and potentially large storage. The volume is specified above in the [[mounts]] section in fly.toml.

• The beefy machine we have specified under [[vm]] makes possible to synchronize faster with the network. So far I've running my node for 4 days and it has checked 38423066/43466716 blocks, which is around the 88%. The accumulated cost is $35.

• The settings auto_stop_machines = 'off' and min_machines_running = 1 are needed because Fly.io will try to stop the machine if there's no incoming traffic to the configured HTTP port.