Defect HC-2282

Search health check API shows subordinate servers as down if repeater is down.

Software

Customer case Applies to Corrected in
CS0092119 V9.0.1.0 9.0.1.12

Observed behavior

The search health check API (/search/admin/resources/health/status?type=container) is used to regularly check the status of the search container to ensure it is in a healthy state. If the search repeater is brought down or is otherwise unavailable, all subordinate search servers are marked as down, and will remain that way until the search repeater is restarted.

Expected behavior

The search service can continue to work even if the repeater is not available. The health check saying that the secondary servers are not available gives the impression that everything is down.

Resolution

This fix introduces an optional mode parameter for type=container. The two valid values for mode are startup and liveness. The default value for mode is startup, which is the current behavior. When the mode parameter has the value of liveness, for solr secondary servers, the health check will return OK if the secondary servers are live.

Once this fix is applied, if the repeater is down, both of these calls will return with a status of 500:

/search/admin/resources/health/status?type=container and /search/admin/resources/health/status?type=container&mode=startup

This call will return a status of 200 if the reapeater is down but subordinate servers are up:

/search/admin/resources/health/status?type=container&mode=liveness