ramkrish86 commented on a change in pull request #1552:
URL: https://github.com/apache/hbase/pull/1552#discussion_r441333721
##########
File path:
hbase-hadoop-compat/src/main/java/org/apache/hadoop/hbase/regionserver/MetricsRegionSourceImpl.java
##########
@@ -302,6 +308,24 @@ void snapshot(MetricsRecordBuilder mrb, boolean ignored) {
regionNamePrefix + MetricsRegionSource.MAX_FLUSH_QUEUE_SIZE,
MetricsRegionSource.MAX_FLUSH_QUEUE_DESC),
this.regionWrapper.getMaxFlushQueueSize());
+ addCounter(mrb, this.regionWrapper.getMemstoreOnlyRowReadsCount(),
+ MetricsRegionSource.ROW_READS_ONLY_ON_MEMSTORE,
+ MetricsRegionSource.ROW_READS_ONLY_ON_MEMSTORE_DESC);
+ addCounter(mrb, this.regionWrapper.getMixedRowReadsCount(),
+ MetricsRegionSource.MIXED_ROW_READS,
+ MetricsRegionSource.MIXED_ROW_READS_ON_STORE_DESC);
Review comment:
Are asking in terms of CPU that we add on while collecting the metric?
Ifyou see we do collect the metric at the HStore level per row when the
StoreScanner completes a row process. That is now a longadder. Seems it is more
performant than AtomicLong. Also the above change that we have done at the
region level is nothing but just get that metric when that runnable thread
keeps running. We don do any metric collection at this level. Are you still
thinking it may be a problem. @saintstack ? BTW thanks for your review here.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
