deploy-specific

name	deploy-specific
description	Deploy one or more explicitly named Zakura testnet or mainnet nodes through GitHub Actions. Use when the user asks to deploy specific nodes or provides a node list. Use deploy-all for a complete fleet rollout.

Deploy Specific Zakura Nodes

Deploy only the nodes explicitly named by the user. One or many nodes are allowed. Never interpret an omitted node list as the whole fleet, and never leave the workflow's node input blank.

For requests to deploy every node, use deploy-all, which applies the fixed testnet-first, half-by-half rollout strategy.

Inputs

Require or confirm:

• ref: Git branch, tag, or SHA to build and deploy.
• nodes: one or more exact names from the inventories below.
• network: required when a node name is ambiguous, such as zakura-compat.
• force_rebuild: optional; defaults to false.
• no_restart: optional; defaults to false.

For any mainnet node, confirm the ref and complete node list explicitly with the user immediately before dispatch.

Node inventory

Testnet:

• zakura-testnet-1
• zakura-testnet-2
• zakura-testnet-3
• zakura-testnet-eu
• zakura-testnet-as
• zakura-compat

Mainnet:

• asia-0
• us-0
• us-east-0
• us-west-0
• canada-0
• europe-west-0
• europe-central-0
• asia-south-0
• asia-pacific-0
• zakura-compat
• zakura-compat-docker

Reject unknown names rather than guessing. If zakura-compat is requested without a network, ask which network.

Preflight

From the repository root:

git fetch origin <ref>
REF=$(git rev-parse FETCH_HEAD)
gh workflow view zakura-testnet-deploy.yml --repo zakura-core/zakura
gh workflow view zakura-mainnet-deploy.yml --repo zakura-core/zakura

Use the resolved full commit SHA for every requested node.

Dispatch

The workflows accept one node per run. For multiple nodes, dispatch and verify each explicit node in order so a failure stops the remaining deployment:

deploy_specific() {
  local workflow=$1
  shift
  local run_url
  local run_id

  test "$#" -gt 0

  for node in "$@"; do
    run_url=$(gh workflow run "$workflow" \
      --repo zakura-core/zakura \
      -f ref="$REF" \
      -f node="$node")
    test -n "$run_url"
    run_id="${run_url##*/}"
    gh run watch "$run_id" \
      --repo zakura-core/zakura \
      --exit-status
  done
}

Examples:

deploy_specific zakura-testnet-deploy.yml \
  zakura-testnet-1 zakura-testnet-eu

deploy_specific zakura-mainnet-deploy.yml \
  us-east-0 europe-central-0

When a request contains both networks, dispatch and verify the explicitly named testnet nodes before requesting mainnet confirmation and dispatching the named mainnet nodes.

Append these workflow fields only when requested:

-f force_rebuild=true
-f no_restart=true

Verification

• Confirm every workflow run succeeded and its final node status is healthy.
• Confirm each node reports the expected commit/version.
• For restarted nodes, confirm RPC height is current and advances.
• For testnet zakura-compat, confirm the zcashd sidecar sync check passes.
• Stop on failure; do not continue to additional requested nodes.
• Report each requested node and its workflow run URL.

Hard guards

• Never dispatch with an empty node input.
• Never add unrequested nodes to the deployment.
• Do not change cache, state, identity, or node configuration paths.
• Only one deploy per network runs at a time (cancel-in-progress: false).

References

• Full-fleet rollout: ../deploy-all/SKILL.md
• Deployer: deploy/deployer/README.md
• Testnet workflow: .github/workflows/zakura-testnet-deploy.yml
• Mainnet workflow: .github/workflows/zakura-mainnet-deploy.yml