Cassandra statistics
Hi All,
What would be the best way to get the following information from Cassandrda ?
Write Requests (rate, count, latency)
Read Requests (rate, count, latency)
Pending Write Operations (flushes, post flushes, write requests, replication of write)
Pending Read Operations (read requests, read repair tasks, compactions)
Pending Cluster Operations (manual repair tasks, gossip tasks, hinterd handoff, internal responses, migrations, misc tasks, request responses)
Compactions
Row/Key Cache (hit ratio, requests count)
Local Writes (rate, count, latency)
Local Reads (rate, count, latency)
SSTable (size, count)
Bloom Filter (space used, false positives ratio)
Gil.
Hi. I am not the expert on Cassandra and therefore forwarded this internal. In the meantime you may however want to check out the following two links
a) http://apmblog.compuware.com/2011/09/20/cassandra-write-performance-a-quick-look-inside/
b) dynaTrace FastPack for Cassandra
Andi
Andi,
Just found out that competitive tools are actually using JMX to collect this data. Can we implement the same easily ?
Gil.
The dynaTrace Fastpack for Cassandra Andi linked above includes many of the JMX metrics with zero-configuration needed:
Yes. dynaTrace has built-in support to capture ANY JMX Metric. Check out the following doc page: Using Custom JMX and PMI Measures in Charts
sorry - how do you get this information from dynatrace? We installed dynatrace on our Cassandra hosts as well as on our application hosts that are making calls to Cassandra, but we are only able to capture basic host level monitoring Is there some configuration that we might have missed?
This is what we have for our sensor placement. We are getting a couple hints; change the placement of sensor packs and NoSQL Cassandra Database hints . not sure what each of these refers to or where these configurations are located. Any help would be great.
Hi
Please download the Cassandra Fastpack that I've linked to: Apache Cassandra Fastpack
Once you have installed that FastPack you have new pre-configured JMX Metrics available in all of your System Profiles. Simply open your System Profile - click on Measures and then on Create. You will find the new measures that Richard showed in the screenshot under Server-Side Performance->Agent based Measures->Cassandra Basics
Andi
I have noticed in the DataStax documentation under "Cluster Metric Keys"
measurements like "read-ops" and "write-ops" I am not able to find similarly named metrics in Dynatrace when trying to create a new measure using custom JMX time measures Is is possibly that Dynatrace is not using the default names that DataStax is using, or is Dynatrace only seeing a fraction of the measures DataStax supports for monitoring Cassandra
Happy to hear your thoughts
Ed
Community Member of the Month
April 2018
Announcing Dynatrace's Community Member of the Month for April 2018, Babar Q.! Click here to read more!