How do you manage Helm deployments for many microservices without duplicating charts?

You can also keep separate application charts while sharing a common library or subchart. Put the reusable templates and defaults in the library chart, then let each service chart provide only its values and any genuinely custom resources. Conditions in the values file can enable optional features without duplicating the core templates.

Avoid forcing every service into one giant chart if their deployment requirements will eventually diverge. A large chart full of conditionals can become harder to maintain than a little duplication. It is also useful to publish shared charts to a chart repository and have services update their chart dependencies through automated pull requests, so changes are reviewed rather than copied manually.

Another option is to create a generic base chart containing the capabilities most teams need, such as deployments, services, ingress, probes, and resource settings. Each application then supplies its own values. This gives teams a consistent deployment model without copying the templates everywhere.

If the services are deployed in mostly the same way, a shared chart with per-service values is a common approach. The templates can iterate over a collection of services, while the values define names, ports, images, replicas, and other differences. This works especially well when a single application release is meant to contain several related services.