Closed
Description
In Nodes stats API ,we can get following io_stats metric:
final long totalOperations;
final long totalReadOperations;
final long totalWriteOperations;
final long totalReadKilobytes;
final long totalWriteKilobytes;
But these metrics do not give an indication of disk busyness,I looked at some metrics, such as iowait, await,ioutil，but they were not rigorous enough, Refer to the collectd and ganglia implementation, They use ioTime to observe disk busyness.
link:https://collectd.org/wiki/index.php/Plugin:Disk
“io_time” - time spent doing I/Os (ms). You can treat this metric as a device load percentage (Value of 1 sec time spent matches 100% of load).
for Elasticsearch , this can be read Field 10 in /proc/diskstats
Would it make sense to add this metric? Thanks!