bharos opened a new pull request, #11790:
URL: https://github.com/apache/gravitino/pull/11790

   ### What changes were proposed in this pull request?
   
   Adds a design doc at `design-docs/gravitino-role-assumption.md` proposing 
**role assumption (SET ROLE)** — letting a caller narrow the active role set 
for a request via an `X-Gravitino-Active-Role` header. The narrowing is 
**subtractive-only** (validated against the caller's roles; can never escalate) 
and applies consistently across access checks, list filtering, and credential 
vending on the native authorization path (Iceberg REST + native Gravitino API).
   
   The doc covers the header grammar, how the server enforces narrowing on the 
existing Casbin model without a model change, transport per engine (Spark works 
today, Trino needs a small setting), security/trust boundaries (cooperative 
header vs. token-bound), prior art (Snowflake/Hive/Postgres), options 
considered, and a phased rollout.
   
   ### Why are the changes needed?
   
   A workload that only needs a few tables still runs with everything its 
identity can reach; misbehavior fails *open* and silent. Role assumption gives 
workloads (and AI agents) real runtime least-privilege and a clean audit 
signal, and restores `SET ROLE` parity for teams migrating Hive 
SQL-Standard-Auth tables to Iceberg behind Gravitino.
   
   This is a **proposal for discussion** (Status: Draft). Key open decisions 
are listed in §12 — most importantly the ownership-vs-narrowing semantics (§5). 
Feedback welcome inline.
   
   Discussion: apache/gravitino#10894
   
   ### Does this PR introduce _any_ user-facing change?
   
   No — documentation only.
   
   ### How was this patch tested?
   
   N/A — design document only.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to