Orchestration and Execution

Overview

Pipelines are new in 4.0. They let you arrange the built-in modules parameter, build, execute command, data transfer, data upload and message push as nodes of a workflow, run it with one click and follow every node and host in a live console. Compared with the fixed flow of application deployment, pipelines are a better fit for:

• multi-step, multi-host release and operations flows with dependencies between steps;
• flows that need input from the operator at run time (version, branch, feature switches);
• artifacts that must be distributed to many hosts followed by further commands;
• results that should be pushed to DingTalk / Feishu / WeCom, or sent as SMS / phone calls through Push Assistant.

Creating a pipeline

1. Click New on the Pipelines page, enter a name and open the editor.
2. The canvas is on the left. Click a node to open the Select Module / Module Config panel on the right; the ⋮ menu of a node lets you add a downstream node or delete the node.
3. Pick a built-in module for every node, fill in its settings and click Save.
4. Click the Run button (lightning icon) in the top-right corner; the list page also has a Run action.

Nodes and conditions

Execution starts at the first node. Every node finishes as success or failure, and the condition of each downstream node decides whether it runs:

Condition	Description
On upstream success	runs only after the upstream node succeeded (default)
On upstream failure	runs only when the upstream node failed; typical for failure notifications or rollbacks
Always	runs regardless of the upstream result; typical for clean-up and result notifications

• The first node has no upstream, so its condition is ignored.
• A node may have several downstream nodes; all that match the condition run in parallel. Give each node only one upstream to avoid triggering it twice.
• Multi-host nodes (execute command, data transfer, data upload) run on the selected hosts in parallel and succeed only when every host succeeds.

Running and the console

When a pipeline contains a parameter node with dynamic parameters, a build node with a Git branch / tag chosen at run time, or a data upload node, a dialog asks for the values or files before the run starts.

• The left side lists every node with its state (waiting, running, succeeded, failed); click a node to see the real-time output of each host.
• Running nodes can be terminated, which sends a termination signal to the hosts.
• Every run is stored as a history entry with an increasing sequence number.

Variables

Variables are injected into node scripts as environment variables:

Source	Variable	Description
Parameter node	_SPUG_<variable>	dynamic parameters (entered at run time) and static parameters (fixed values)
Build node	SPUG_SET_*	variables exported in the build command with the SPUG_SET_ prefix are passed to later nodes without the prefix
Build node	_spug_git_tag / _spug_git_commit	the Git tag / commit chosen at run time
Push nodes	$SPUG_PIPE_NAME, $SPUG_NODE_NAME, $SPUG_STATE_TEXT, $SPUG_DATETIME	pipeline name, node name, upstream state text and current time, available in push content only

Example: export a version number in the build node for later command nodes:

# build command of the build node
mvn -q package
export SPUG_SET_JAR_VERSION=$(cat VERSION)

# a later execute command node
java -jar app-$JAR_VERSION.jar

See Node modules for the settings of every module.

Permissions

Pipeline permissions are under Roles / Feature permissions / Pipelines (view, new, edit, delete, run). The hosts selectable in nodes follow the host permissions of the role.