Skip to main content

Several hosts and outside validators

A deployment can place each validator on its own host, and include validators that someone else runs.

Validators on their own hosts​

Declare the hosts under hosts and place each node:

[env.testnet.sequencer]
host = "a"                         # the sequencer's host also runs the relayer
key = "acme-sequencer"             # signs its requests to the validators

[[env.testnet.validators]]
name = "3"
key = "acme-v3"
host = "b"

[env.testnet.hosts.a]
provider = "ssh"
address = "ops@a.example"
public_url = "https://lane.example"
private_address = "10.0.0.2"       # optional: a private network between the hosts

[env.testnet.hosts.b]
provider = "ssh"
address = "ops@b.example"
public_url = "https://b.example"
private_address = "10.0.0.3"
  • Each host gets only what it runs: its validators' configs, units and keys.
  • Signed requests. Once a validator runs on another host, [sequencer] key is required. The sequencer signs every request it sends to a validator's /v1/sign, and validators answer only those, within 60 seconds of signing.
  • Reaching each other. With a private_address, hosts reach each other's nodes there. Without one, they go through each host's public_url.
  • Moving a validator. Change its host and apply: it starts on the new host, then stops on the old one, and its key leaves.

Validators run by others​

A validator someone else operates is a reference, never deployed:

[[env.testnet.validators]]
name = "partner"
key = "partner-v"                  # stellar keys add partner-v --public-key G…
url = "https://validator.partner.example"

It joins the signer set, and the sequencer asks it to sign. Its operator runs the template's node as a validator with sequencer_key set to your sequencer's public key; ${node.sequencer.key} gives it to them as an output. See Run a validator.

The hosts reference has every rule.