You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It would be great to be able to combine multiple sources of secrets under same item.
I.e. manually created ones + automated ones from different components/modules.
Since there is no access control granularity past vault-level - there is no security downside to this.
However from UX standpoint - this has a benefit of keeping everything in one place vs needing to spam an item per individual piece of automation code + anything manual you'd have to add on top.
Use cases
PostgreSQL users/urls, multiple levels of grafana tokens (cloud/stack), secrets for a single app that comes from multiple sources (db, cdn, deploy keys, buckets, metrics/logs/traces publisher keys).
If you use something like terragrunt or magic python wrapper - you get separately scoped component for each service. (to reduce graph size, apply different concurrency or run less in CI/CD.. etc)
Proposed solution
Have dedicated section/item resource. Possibly needing a "custom composite" item type that supports this.
Is there a workaround to accomplish this today?
Hint of a workaround is to use ignore_changes+protect_from_destroy for manual secrets, but still does not handle multiple sources. Also, other code having to ignore things it should not know about is a scope leak.
The text was updated successfully, but these errors were encountered:
Summary
It would be great to be able to combine multiple sources of secrets under same item.
I.e. manually created ones + automated ones from different components/modules.
Since there is no access control granularity past vault-level - there is no security downside to this.
However from UX standpoint - this has a benefit of keeping everything in one place vs needing to spam an item per individual piece of automation code + anything manual you'd have to add on top.
Use cases
PostgreSQL users/urls, multiple levels of grafana tokens (cloud/stack), secrets for a single app that comes from multiple sources (db, cdn, deploy keys, buckets, metrics/logs/traces publisher keys).
If you use something like terragrunt or magic python wrapper - you get separately scoped component for each service. (to reduce graph size, apply different concurrency or run less in CI/CD.. etc)
Proposed solution
Have dedicated section/item resource. Possibly needing a "custom composite" item type that supports this.
Is there a workaround to accomplish this today?
Hint of a workaround is to use ignore_changes+protect_from_destroy for manual secrets, but still does not handle multiple sources. Also, other code having to ignore things it should not know about is a scope leak.
The text was updated successfully, but these errors were encountered: