clarax commented on a change in pull request #3356:
URL: https://github.com/apache/hbase/pull/3356#discussion_r646861474
##########
File path:
hbase-balancer/src/main/java/org/apache/hadoop/hbase/master/balancer/StochasticLoadBalancer.java
##########
@@ -343,15 +343,15 @@ boolean needsBalance(TableName tableName,
BalancerClusterState cluster) {
sendRejectionReasonToRingBuffer(() -> getBalanceReason(calculatedTotal,
calculatedMultiplier),
costFunctions);
}
- if (LOG.isDebugEnabled()) {
- LOG.debug("{} {}; total cost={}, sum multiplier={}; cost/multiplier to
need a balance is {}",
- balanced ? "Skipping load balancing because balanced" : "We need to
load balance",
- isByTable ? String.format("table (%s)", tableName) : "cluster",
- total, sumMultiplier, minCostNeedBalance);
- if (LOG.isTraceEnabled()) {
- LOG.trace("Balance decision detailed function costs={}",
functionCost());
- }
+ LOG.info("{} {}; total cost={}, sum multiplier={}; cost/multiplier to need
a balance is {}",
Review comment:
Hmm, right. Need to explain cost /multiplier is the measurement how
balanced the cluster for balancer at a scale on [0,1] 0 being the perfectly
balanced and 1 is the worst it could be. But the actual value still doesn't
seem very intuitive to operator. We can explain if we demand the max of
absolute number of unbalanced nodes to be constant, the bigger the cluster, the
smaller of this value would be.
--
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:
[email protected]
