Data collection behavior for network devices in the high availability setup for On-Premise Pollers

Data collection behavior for network devices in the high availability setup for On-Premise Pollers

When high availability (HA) is configured for On-Premise Pollers in Site24x7, it ensures continued data collection during a primary On-Premise Poller failure. However, some differences in data collection behavior may be observed when the secondary On-Premise Poller takes over. This article outlines the expected behavior and explains why only certain types of data may be collected in failover scenarios.

Behavior during normal operations

When the primary On-Premise Poller is operational:

  • The primary On-Premise Poller handles all SNMP-based data collection (such as CPU, memory, and interface metrics).
  • It also collects availability data, such as the RTT, packet loss, and response time.
  • Monitoring data is synced with the secondary On-Premise Poller every six hours.
Notes

HA is supported only for network availability and performance monitoring and agentless server monitoring.

Behavior during failovers

When the primary On-Premise Poller goes down, and the secondary On-Premise Poller becomes active:

  • The system automatically switches monitoring responsibilities to the secondary On-Premise Poller.
  • Availability data (such as ping and packet loss) is collected immediately.
  • However, SNMP-based metrics (such as CPU, memory, and interface statistics) may not be collected immediately, or may not be collected at all for certain devices.

Reasons for missing SNMP data after failovers

  1. Status checks first: Upon a failover, the system performs a status check for all devices before beginning full data collection. This ensures that availability monitoring for all the monitors resumes immediately.
  2. Gradual initialization: After switching to the secondary On-Premise Poller, data collection will resume gradually for all monitors, especially when a large number of devices is being monitored.
  3. A monitor synchronization delay: Monitors are synchronized from the primary On-Premise Poller to the secondary On-Premise Poller every six hours. If a device was recently added to the primary On-Premise Poller, SNMP data collection will not occur until the next synchronization.

What you can expect

  • For already-synced devices:
    • Availability data collection resumes immediately.
    • SNMP-based data collection begins shortly after the status checks.
  • For recently-added devices (not yet synced):
    • Availability checks may begin, but full SNMP data collection could be delayed until after the next sync cycle or manual intervention.

Recommendation

  • Contact the Site24x7 support team if SNMP data collection does not resume as expected for any particular device.