Details how Crosswork Data Gateway pools provide high availability through active and standby instances, including pool protection states, failover behavior, error detection, workload reassignment, and recovery procedures to minimize disruption during gateway failures.
A Data Gateway high availability pool is a group of Crosswork Data Gateway instances that
-
provides device-specific data collection with minimal disruption
-
enables high availability by supporting active and standby roles, and
-
automatically assigns workloads to standby instances when a failure is detected.
A Data Gateway pool can be in one of several states:
-
Protected: All instances are UP, with matched pairs of active and standby
-
Not protected: All standby instances are DOWN; none are available to replace an active
-
Limited protection: At least one standby instance is UP
-
None planned: No standby instances were configured
Understanding Data Gateway failover in an HA environment
CDG1 (active), which has a southbound IP address, becomes unresponsive due to port failures or cable disconnections. The Crosswork Network Controller detects this outage and activates CDG2 (standby) to replace CDG1. At that point, CDG1 and its replacement share the same device-facing IP address. Therefore, you must power off any failed Data Gateway (using VMware) to avoid conflicts. Only power it back on after the issue causing unresponsiveness is resolved and the gateway can rejoin the pool.
Handling Data Gateway errors and recovery
The Data Gateway manager conducts liveliness checks every 10 seconds; after six missed checks (~60 seconds), a Data Gateway is set to ERROR. If a Data Gateway in a protected pool enters ERROR, devices and jobs are reassigned to a standby instance, ensuring continuity. When a failed instance recovers, it rejoins the pool as standby.
If the Data Gateway identifies interface connectivity issues for northbound communication as part of its health status, it may also respond to the liveliness check and report an ERROR state.
The Data Gateway manager checks the Operational State of the Data Gateway every 20 seconds. When the active instance is in the ERROR state, the Data Gateway manager initiates a failover, resulting in a spare instance from the pool becoming the new active instance.