We don't currently provide state-specific JMX metrics, with the exception of
the ERROR state (that is ErrorPartitionGauge, and the total number of
partitions is PartitionGauge). There's also a DifferenceWithIdealState gauge,
but, looking at the code, that seems to only be correct for CUSTOMIZED
rebalance mode.
If we fix DifferenceWithIdealState to work with any rebalance mode, it will
provide exactly what Hang needs.
Kanak
Date: Mon, 2 Jun 2014 10:57:23 -0700
Subject: JMX metrics
From: [email protected]
To: [email protected]; [email protected]
Hi,
Saw this question on IRC,
Hi, this is Hang. Just trying apache helix, it works as expectedly. And I have
a question, how do I monitor how many partitions are in each state, e.g., say
we are using OnlineOffline state model, how do I monitor how many partitions
are in Online state, and how many partitions are in Offline state. I know we
can listResourceInfo of a cluster via helix-admin, is there any JMX metrics for
it?
9:42 the version we are using is 0.6.3
We do have JMX beans, do we support this ?
thanks,
Kishore G
