Some days back one of the servers went down but we don't know at that time, we get to know after some days, we don't have logs to check how it happens, so for this what is the query or solution or alerts for this case.
Does the server send a heartbeat message? If you are looking for specific hosts then;
If yes, something of this logic. And put it as a savedsearch and run it every x minutes for alerting to your mail or Alerting System
index=someindex sourcetype=someheartBeat | stats count | eval AlertFlag=if(count > 0, "No","Yes")
Or you can based the logic on internal index every 30mins or so.
index=_internal sourcetype=splunkd source=*metrics.log host=someimportantHost| stats count | eval AlertFlag=if(count > 0, "No","Yes")
..
@phanichintha
Try
|metadata type=hosts index=_* index=*
|where now()-lastTime > 10
Finds the difference between now and last time the host reported an event to any index and alerts if the difference is greater than 10 seconds. You can adjust the threshold according to your requirement (in secs)