opscenter | Installs DataStax OpsCenter

DataStax Agent writes collected metrics directly, not sending them to the OpsCenter itself, so it doesn't depend if OpsCenter itself is setup in failover mode, or not.

By default, agent should use LOCAL_ONE or ONE (don't remember), and it isn't configurable. But you just need to remember that consistency level is how many nodes should confirm successful write - writes themselves are are sent to all replicas, so even if node that confirmed write crashed, then there is still a chance that data is written to other replicas.

You need to make sure that you have latest definitions downloaded.

Your file platforms.edn (in /var/lib/opscenter/definitions in package install) should contain following information:

You need to overwrite the environmental variable OPSC_JVM_OPTS. The option you need to look at is Xmx (Maximum Heap Space)

Tarball Installation:

You can use Grafana and add prometheus as a data source. Once done, you should be able to use PromQL for graphing the widgets in a dashboard.

You need to use the metric names depending on how you have mapped in the Prometheus config yaml.

My config has following mapping for node level metrics:

• pattern: org.apache.cassandra.metrics<>(Count|OneMinuteRate|FiveMinuteRate)
• name: cassandra_$1_$2_$3_$4

So for read requests (& write requests) per sec, I am using the metric - cassandra_ClientRequest_Read_Latency_OneMinuteRate - for plotting the oneminuterate.

"In Studio, under the run arrow, select Execute using analytic engine (Spark) before running." The drop-down is "real-time" by default, you should switch it to "Analytic" https://docs.datastax.com/en/dse/6.0/dse-dev/datastax_enterprise/graph/quickStart/graphQSTOC.html

For Graph the only option is to create necessary indexes, then get the existing schema.xml via dsetool get_core_schema, do modifications, and load it again with dsetool reload_core - the reloading of core should be done in every data center, but not on every machine of the data center...

Don't forget, that before reloading the core with modified schema, you need to upload synonym files with dsetool write_resource.

See the DSE Search documentation for full list of the options for dsetool command.

You need to provide following settings as described in documentation:

• -u - to specify user name
• -p - to specify password
• --driver.auth.provider DsePlainTextAuthProvider - to select corresponding authentication provider.

So, in the end we managed to get this working.

An update had somehow activated 'Force authentication' resulting in 401 between Jenkins projects and sonarqube

We disabled force authentication so Jenkins can get the status from the sonarqube scans, and this resolved the issue.

TL;DR: No. OpsCenter is a closed source product, which was not tested with Scylla. Part of it that uses Apache Cassandra CQL and JMX will probably work, while others might not. In addition to the open source, Scylla monitoring stack (base on Prometheus and Grafana), ScyllaDB has its own close version product for cluster management named Scylla Manager.

Tzach (Scylla Product Manager)