This allows to create a docker stack with all the containers needed for local testing/development like grafana, cassandra, promethus, tlp-stress. gradlew preview to run the entire monitoring stack. Run locally for development or tests purposes Then choose the name of the dashboard and folder where it should be stored.Īt this stage, the dashboard should be available and you should have been redirected to it.Īny problem raises? Don't hesitate to create an issue or a pull request to fix the documentation.Download and import the JSON files from the Grafana UI.Grafana dashboards in JSON format are here or here. It is very handy to build dashboards or to debug a setup that is not working as expected. This expression browser allows you to see what Prometheus exposes. Other configurations of prometheus are possible, as described in Prometheus documentationĪt this stage, you should be able to see Operating System and Cassandra metrics querying the prometheus expression Browser.Replacement: '$' # Add datacenter suffix if any # Filters are more performant than relabel_configs but require Prometheus 2.3+ Add jmx_exporter jar file to /usr/share/cassandra/lib/jmx_prometheus_javaagent-0.12.0.jar.On all the nodes, take the following steps. This could change in the future but for now we aim at building on top of this exporter, and using default naming (no renaming, yet). After distinct attempts with a few distinct exporters, we decided to go with the official one, recommended in Prometheus documentation for collecting JMX metrics (Cassandra, or Kafka are explicitly stated there). Export Cassandra Metrics (through JMX)įor cassandra metrics and when using Prometheus, there is a lot of distinct metrics exporters you could use out there. Newer (and older) versions of those, might also work well, with no action or minor tweaks. Runbook - How to Use the TLP dashboards in my Cassandra cluster(s)? Prerequisite That being said, we believe the existing charts are probably a good starting point to get an accurate feel for your Cassandra clusters. We will handles those request with the highest priority. Please send a feedback or a pull request with the fix and make it available to everyone. Some charts might be inaccurate or even wrong.For now metrics are not filtered out, which on bigger cluster or with a lot of table could be a problem).We are building more dashboards and improving the existing ones, that should not require any big modification on operators side once dashboards are already set up. This should allow you to use a 'TLP Cassandra Overview' dashboard, which is the critical dashboard to detect anomalies for a Cassandra cluster. Cassandra Dashboards based on Prometheus and Grafana
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |