Kubernetes plugin - best practices for working with multiple clusters

[mynameiswhm]

At @avito-tech we have tens of Kubernetes clusters already (and growing), so when I tried using Kubernetes plugin for querying what we have, I quickly realized that it's really inconvenient to query across multiple clusters — every Kubernetes context is in different connection.

But looking, for example, at AWS plugin, there's region field for any major entity (along with regions configuration option) — I can query everything and group/filter based on region.

I've looked at the plugins available right now, and most of them are either single-instance or SaaS-like (where you already have single or few accounts, like AWS), but Kubernetes case is quite different from that.

What's the best practices for working with "multiple instances" setups like the one that we have?

[e-gineer]

The recommended best practice is to:

• Create a connection per Kubernetes cluster.
• Setup an aggregator, so you can query across all those connections at the same time.

Aggregator docs - https://steampipe.io/docs/using-steampipe/managing-connections#using-aggregators

connection "k8s_all" {
  type        = "aggregator"
  plugin      = "kubernetes"
  # Named connections
  connections = ["k8s_conn_a", "k8s_conn_b"]
  # Or use wildcards
  # connections = ["k8s_conn_*"]
}

Each table in the kubernetes plugin then has a context_name column so you can work out which cluster / connection the results are coming from.

Does that help?

[mynameiswhm]

Thanks, looks like a very important part of documentation that I've completely missed! Works like a charm!