Hello shashikrishnak,
I wouldn't recommend installing Telegraf directly on the Aria Operations for Logs appliance just to monitor Cassandra. Aria Operations for Logs is an appliance, and adding third-party agents/packages to it isn't the approach I'd use for service-level monitoring.
For Cassandra, the most useful health check is the Cassandra node state. Broadcom uses the following command in its own troubleshooting procedures:
/usr/lib/loginsight/application/lib/apache-cassandra-*/bin/nodetool-no-pass status
A healthy node should normally show UN (Up/Normal). Broadcom specifically recommends checking this across all Aria Operations for Logs nodes when troubleshooting Cassandra availability.
The challenge is getting that internal Cassandra state into Aria Operations. I don't believe there is a native Aria Operations metric that directly represents the internal Cassandra service health of an Aria Operations for Logs appliance.
I would therefore avoid trying to monitor the Cassandra process itself with an agent. Instead, I'd monitor the appliance/cluster using supported health indicators and create alerts around conditions that commonly accompany Cassandra failure, such as:
- Aria Operations for Logs node availability
- appliance CPU/memory
- /storage/core and /storage/var utilization
- cluster/node connectivity
- Log Insight service availability
Disk utilization is particularly important. Broadcom documents cases where full /storage/var or /storage/core filesystems result in Cassandra becoming unavailable and nodes becoming disconnected or the UI becoming inaccessible.
If your requirement is specifically "alert in Aria Operations when Cassandra is Down/Not Normal", you would likely need a custom monitoring mechanism that periodically evaluates nodetool-no-pass status and publishes that result into Aria Operations as a custom metric/event, rather than installing Telegraf inside the Logs appliance.
I would also be cautious about automatically restarting Cassandra based on such an alert. Cassandra being down can be a symptom of disk exhaustion, excessive ingestion, corruption, or another cluster problem. Broadcom recommends contacting Support when a node doesn't return to UN rather than treating a Cassandra restart as the universal fix.
So, in short: Telegraf on the Aria Operations for Logs appliance isn't the route I'd take. Monitor the supported appliance/cluster health indicators, or use an external custom check if Cassandra-specific UN/DN status is mandatory.
Hope this helps.