Add a Groundplex

A Groundplex consists of one or more nodes. Each node is the Snaplex software running on a virtual or physical machine. For development, you can use a single-node Groundplex. In production, use multiple nodes so that:

• Pipeline execution load spreads across the nodes.
• Version upgrades roll out one node at a time, with no downtime.
• The Snaplex keeps accepting new pipeline executions if one node fails or restarts for maintenance, though pipelines already running on that node fail.

Nodes can be of two types:

• The default JCC node runs pipelines.
• A FeedMaster node distributes Ultra Task requests across JCC nodes. To make a node a FeedMaster, map its hostname to the FeedMaster role in the Snaplex node types field on the Node properties tab.

To create a new Groundplex:

First, configure the Snaplex in Admin Manager. You set its name and identity, its logging, how its nodes behave, and whether they use a proxy — each on its own tab. Saving creates the Snaplex and adds a Downloads tab.

Second, install it. The Downloads tab provides a software image for your platform — download it for Linux or Windows, or copy it for Docker — and a configuration file (.slpropz) holding your saved settings. Install the image on each machine, then apply the configuration file. A machine without the configuration file doesn't join the Snaplex.

If you invoke Triggered, Instant, or Ultra Tasks directly via REST API calls to your nodes, you also need a load balancer in front of them so requests reach the right one. Pipeline executions invoked through the control plane, Headless Ultra Tasks, scheduled pipelines, and child pipelines run via the Pipeline Execute Snap don't need one. APIM requires a load balancer.

You configure a Groundplex on four tabs:

Settings

What identifies this Snaplex and binds its nodes to it, which project it belongs to, which Snaplex version the nodes run, and how task requests reach them.

Logging

How much detail your nodes log and how long they keep it before rotating and deleting old log files.

Node properties

How much execution capacity each node has (slots consumed per pipeline based on its Snap count, memory limits, and capacity reserved for interactive Designer executions) and how nodes behave during restarts, version upgrades, or maintenance.

Node proxies

Whether outbound traffic from your nodes goes through your organization's proxy server.

By default, every node in a Groundplex runs as a JCC (the pipeline execution service). You only need to override this if you want a specific machine to run as a FeedMaster instead — the node type that manages request-and-response messaging for Ultra Pipelines.