Hi
I read your docs and I still have some gaps
- How to implement it in cluster mode?(Helm on EKS).
- How to specify which metric I want to export? (I need only jobs, memory, RPC, and threads).
Thank you
http://docs.dremio.com/advanced-administration/monitoring/
dremio-env File on Coordinator Node
DREMIO_JAVA_SERVER_EXTRA_OPTS='
-Dcom.sun.management.jmxremote.port=9048
-Dcom.sun.management.jmxremote.authenticate=false
-Dcom.sun.management.jmxremote.ssl=false'
dremio-telemetry.yaml file add the following to export
# Dremio telemetry configuration file
# Control if this file should automatically be reloaded, and at which interval.
auto-reload:
enabled: True
period: 30
unit: SECONDS
metrics:
- name: reporter185_jmx
reporter:
type: jmx
# includes: []
# excludes: []
ReStart Dremio
Test JMX Dump
Google jmx-dump
Download standalone binary or jar from latest release from GitHub - r4um/jmx-dump: Dump JMX Metrics as JSON on command line.
java -jar ./jmx-dump-0.10.3-standalone.jar -p 9048 --dump-all -h 172.25.0.185
Verifies that you have all the Dremio JMX Metrics available on PORT 9048
The link Dremio lists the JMX metrics you can track on the Coordinator.
For Helm Deployment you will need to add the template config files.
**Note: ** You could do the same on Executor nodes to track the the JMX metrics for Executors as listed on the link shared above.
Alternately a simpler approach use SQL API to execute Queries against sys.nodes and sys.memory tables to get CPU and Memory Usage Statistics