Add the cache status field to the Traffic Monitor CrStates endpoint

[jrushford]

This Feature Request affects these Traffic Control components:

• Traffic Monitor

Description

Currently the tc-health-client is using the Traffic Monitor api/cache-statuses endpoint to find and mark down unhealthy parents. This endpoint it turns out is for local TM state and may not reflect the state as reached by consensus of all TM's. The health-client is being switched to use the publish/CrStates endpoint instead.

However, the publish/CrStates endpoint lacks the status field that is available in the api/cache-statuses endpoint. This field was logged by the health-client so that it may be understood why the parent was marked down ie, the health threshold that triggered the markdown. It would be invaluable to continue this logging if the field could be added to the publish/CrStates endpoint or a query parameter that would let the client ask for the field as part of the response. Querying both endpoints is not desirable.

Also, a date time field should be added to the publish/Crstates endpoint response that shows when the health monitor last polled the cache. This could also be added to the response only when requested with a query parameter.

Example status information from api/cache-statuses:

• REPORTED - availableBandwidthInKbps too low (0.00 < 75000000.00
• REPORTED - available
• REPORTED - queryTime too high (1520.00 > 1500.00)

Possible new struct layout, see lib/go-tc/crstats.go:

type IsAvailable struct  {
        IsAvailable bool  
        Ipv4Available bool
        Ipv6Available bool   
        Status string // new field status and health threshold reason when unavailable
        LastHmPoll time.Time // new field last Health monitor poll` 
}