Summary

Current implementation of services was designed long time ago, it is rather limited and has significant limitations. Major limitations are:

• scalability: it does not scale well, also bad user experience when number of services exceeds few hundreds
• hard to configure: mapping based on triggers does not work well and requires significant effort
• very limited visualization: no graphical representation of services, no good reporting
• limited SLA calculation rules
• no alerting in case of service status changes
• user permissions: all or nothing

Proposed functionality will introduce alerting for services.

Acceptance

Current implementation of Services must be extended to support alerting for service status changes:

1. Each service status change will generate a new event (new event source EVENT_SOURCE_SERVICES)
2. Zabbix will be extended to allow alerting on the service events similarly to existing functionality of trigger-based actions
   1. Normal, recovery and update operations must be supported
   2. Escalation options must be supported
   3. The following conditions will be supported:
      1. Service tag name
      2. Service tag value
      3. Service status:
         1. Condition: equals, does not equal, is greater than or equal, is less than or equal
         2. Service status: OK, Not classified, Information, Warning, Average, High, Disaster
      4. Service name (equals or contains)
      5. Service (multi-select of services)
3. A new set of macros must be introduced:

1. 1. {SERVICE.NAME}

      : service name

   2. {SERVICE.STATUS}

      : service status in text

   3. {SERVICE.NSTATUS}

      : service status numeric

   4. {SERVICE.TAGS}

      : a comma separated list of service tags

   5. {SERVICE.TAGSJSON}

      : a JSON array containing service tags

   6. {SERVICE.TAGS.<tag name>}

      : tag value

   7. {SERVICE.ROOTCAUSE}

      : list of problem events that caused service to fail sorted by severity and host name (host name, event name, severity, age, tags)

      1. Only problems having real impact on the service must be in the list 
   8. No other macros will be supported to keep this functionality "disconnected" from Zabbix core functionality
2. Media type's message template must be extended with new message types "Service", "Service recovery" and "Service update"
3. Only users having at least read access to services will be notified