This can be a leading indicator of a potential loss of service or flood of requests.
This is measured by: # of Flushes / Time spent on Flushes (ms) for the evaluated time window.
Increased Flush Latency is an indication that your disks cannot keep up with your cluster's demands and you may need to reconfigure the 'flush_threshold_size' setting to reduce the translog size needed to trigger a flush operation.
This is measured by: # of Docs Indexed / Time spent Indexing (ms) for the evaluated time window.
Increased Flush Latency is an indication that you are trying to index too many documents at one time and you may need to reconfigured your settings to increase performance.
This quickstart includes dashboards and alerts for popular signals regarding Elasticsearch cluster health and performance. Use this quickstart together with the New Relic Elasticsearch On Host Integration and New Relic Infrastructure agent log forwarding to get insight into the performance of your Elasticsearch clusters.