Feature Request

Collect metrics at the node level instead of the cluster level.

Proposal:

Couchbase doesn't recommend collecting and using cluster wide metrics for monitoring as they are already aggregated by the cluster. Instead, they recommend to monitor each node individually. This is what their Prometheus exporter does. See https://github.com/couchbaselabs/cbprometheus_python.

On page 14 of their monitoring guide says this for bucket stats:

Cluster-Wide Individual Bucket Stats
These endpoints are informational and should not be used for monitoring as they are an aggregate for the
entire and cluster and the best practice is to monitor each node individually.

• Insecure: http://localhost:8091/pools/default/buckets/{BUCKET}/stats
• Secure: https://localhost:18091/pools/default/buckets/{BUCKET}/stats

Node-Level Individual Bucket Stats
Each node in the cluster running the data service should be monitoring individually using the endpoint listed
below.

• Insecure: http://localhost:8091/pools/default/buckets/{BUCKET}/nodes/{NODE}/stats
• Secure: https://localhost:18091/pools/default/buckets/{BUCKET}/nodes/{NODE}/stats

Current behavior:

Metrics are being collected at the cluster level which is not best practice.

Desired behavior:

Metrics should be collected at the node level.

Use case:

According to their monitoring guide, their sample monitoring scripts, and conversations I've had with their architects, they only recommend collecting metrics at the node level to get a more accurate view of the cluster.

I do want to thank you for adding this into Telegraf. Couchbase doesn't get enough love and I am thrilled to see it being supported by the community.