Deploy FeedMaster node
A FeedMaster is a specialized type of Snaplex node that manages the distribution of Ultra Task requests across the JCC nodes within the Snaplex. It operates as a Java Virtual Machine (JVM) acting as an intermediary between the JCC nodes (which execute the pipelines) and the client.
- The FeedMaster controls and distributes Ultra Task requests to ensure efficient load balancing and task execution across JCC nodes.
- In a Cloudplex, FeedMaster nodes are deployed by SnapLogic.
- In a Groundplex (self-managed Snaplex), your organization sets up and manages the FeedMaster nodes.
A FeedMaster node runs on a Groundplex host and has the same hardware, operating system, and network requirements as any JCC (execution) node. Before you deploy, confirm the following:
- To use a FeedMaster node for Ultra Task pipelines, your environment (Org) must be subscribed to the Ultra feature.
- You must have a Groundplex. For Cloudplex FeedMasters, contact your CSM.
- The host meets the same hardware, operating system, and network requirements as a JCC node. See Groundplex requirements: Hardware and operating systems and Groundplex requirements: Network.
- Because a FeedMaster queues Ultra Task requests on disk, it needs additional
storage beyond the JCC disk requirement. The default is 10 GB, configurable with
the
broker.disk_limitproperty. See Deployment architecture for Ultra Tasks. - Install the FeedMaster node physically near the Snaplex instances and the data the pipeline passes, to support low-latency processing.
The FeedMaster node uses several dedicated ports: the FeedMaster HTTPS port (default
8084), the embedded message broker port for Ultra Tasks (default
8089), and the FeedMaster HTTP port (default
8091). For the complete port list and firewall requirements, see
Groundplex requirements: Network. If other
resources use these ports, you can change the port number on the Node
Properties tab.
A node determines its server type from its local configuration when the node process starts, before it registers with the SnapLogic Control Plane. If you start a node before it has been designated as a FeedMaster, it comes up as a regular JCC node and can begin receiving pipeline executions. Designate the FeedMaster role before you start the node for the first time (see the steps that follow).
For details about server requirements, see Install a Groundplex on Linux for Groundplex installation and Groundplex Deployment Planning for specific configurations.
/healthz URL to check the health status of each FeedMaster. For load balancer requirements, see Groundplex
requirements: Load balancer.Troubleshooting
- The node starts as a regular JCC node instead of a FeedMaster. The node
was started before the FeedMaster server type was applied to its local
configuration. Apply the server type before the first start, as described in the
Server type step above. For nodes still configured through
global.properties, see Node migration. - The FeedMaster rejects Ultra Task requests with
HTTP 503 Service Unavailableand a message such asRequest storage is low, try again later
. The message-spool disk is at or near its limit. Free disk space or increasebroker.disk_limit. See Deployment architecture for Ultra Tasks. - A restarting FeedMaster keeps receiving requests. The load balancer is not
health-checking the
/healthzURL, so it continues routing requests to a node that is offline or restarting. See Monitor Snaplex health and Groundplex requirements: Load balancer. - You cannot tell which node is the FeedMaster. In the Monitor application, FeedMaster and execution (JCC) nodes use different icons. See Monitor FAQs.
- You need to investigate FeedMaster errors. Check the FeedMaster log files
(
feed_master.json,feed_master_error.json, and related logs). See Snaplex logs.