[ https://issues.apache.org/jira/browse/HBASE-18469?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16106779#comment-16106779 ]
Yu Li commented on HBASE-18469: ------------------------------- [~zhangshibin] Thanks for filing this JIRA and let me try to explain. The difference mainly comes from scan with caching. Each scan request will be counted as one in {{RsRpcServices#requestCount}} no matter with caching or not, while the {{readRequestCount}} is counted inside {{HRegion}} and will increment every time {{next(List<Cell> outResults, ScannerContext scannerContext)}} is called. So if scan with caching 10, then readRequestCount will be counted 9 more than requestCount, thus finally totalRequestCount doesn't equal to the sum of readRequestCount and writeRequestCount. Considering how we're dealing with the {{multi}} call (we will increase requestCount per region action), I think it makes sense to also count caching in for {{RsRpcServices#requestCount}}. And if we all agree on this, I could supply a patch (should be straight forward). Thanks. cc more people for more suggestions (besides those already watching) [~yangzhe1991] [~Apache9] [~busbey] [~ghelmling] [~anoop.hbase] [~stack] > Correct RegionServer metric of totalRequestCount > -------------------------------------------------- > > Key: HBASE-18469 > URL: https://issues.apache.org/jira/browse/HBASE-18469 > Project: HBase > Issue Type: Bug > Affects Versions: 1.2.0 > Reporter: Shibin Zhang > Priority: Critical > > when i get the metric ,i found this three metric may be have some error as > follow : > "totalRequestCount" : 17541, > "readRequestCount" : 17483, > "writeRequestCount" : 1633, -- This message was sent by Atlassian JIRA (v6.4.14#64029)