Step 7: Determine Service Availability

By default, Horizon monitors a number of services including ICMP, DNS, Elasticsearch, SMTP, FTP, SNMP, HTTP(S), SSH, Postgres, IMAP, and POP3. Using a service monitor framework with pollerd as the main component, Horizon detects if a service is available, then monitors it for its availability and response time. You can view a summary of this availability information, calculated over the last 24 hours, on the main dashboard. You can also view a summary of availability information on a per-node basis.

Figure 1. Service availability on the node detail page

Pollerd schedules and runs service monitors, as defined in poller-configuration.xml. When a service monitor detects an outage, pollerd sends an event which Horizon uses to create an alarm. You can also use events to generate a notification for on-call network or server administrators.

Beyond Quick Start

Refer to Service Assurance in the Deep Dive section for more information, including ways to customize service monitoring and minimize notifications:

Modify poller-configuration.xml to change pollerd behavior.
Group service configurations into polling packages.
Use path outages to suppress notifications based on node interdependence.
Detect short service outages using a downtime model.