We provide some general alarm configuration suggestions for your reference according to our realistic experiences.
|Indicator||Suggested Alarm Threshold||Description|
|Cluster Status||An alarm is given, when the monitoring value is >=2 for 10 consecutive times (i.e., in red)||0 means green, the cluster is in the healthiest status; 1 means yellow, the high availability of the cluster is affected, but the search results are still complete; 2 means red, the cluster is abnormal, the search can only return part of the data|
|Node Disk Util||An alarm is given when the maximum value is >=75 for 10 consecutive times (more than 80% prohibited)||Too high disk utilization will cause the failure of data being written normally|
|Node HeapMemory Util||An alarm is given when the maximum value is >=85 for 10 consecutive times (more than 90% prohibited)||Search performance will be affected when the value is too high|
|Node CPU Util||An alarm is given when the maximum value is >=95 for 10 consecutive times||If the CPU use rate is too high, it will decrease the cluster node processing capacity or even cause system breakdown|