HADOOP-19940. [ABFS] Add no op support for set permission for RBAC and HNS enabled#8607
Open
anmolanmol1234 wants to merge 2 commits into
Open
HADOOP-19940. [ABFS] Add no op support for set permission for RBAC and HNS enabled#8607anmolanmol1234 wants to merge 2 commits into
anmolanmol1234 wants to merge 2 commits into
Conversation
Contributor
Author
============================================================
|
anmolanmol1234
changed the title
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Customers running Spark and Hadoop workloads on ADLS Gen2 with an RBAC-only authorization model (no ACLs assigned, all data access governed by Azure RBAC role assignments) are frequently blocked by framework-generated setPermission() calls.
setPermission() is invoked implicitly by:
Hadoop commit protocols (v1 and v2)
Spark's FileOutputCommitter on job commit
distcp when preserving permissions
Hive/Impala when writing partitions
Various other job-orchestration paths
On HNS-enabled accounts, setPermission() translates to a SetAccessControl request on the ADLS Gen2 REST surface, which requires ACL-management permissions (Microsoft.Storage/storageAccounts/blobServices/containers/blobs/manageOwnership/action or equivalent via a POSIX ACL entry granting the caller ACL modification rights).
In RBAC-only deployments, callers typically have full data-plane RBAC roles (e.g. Storage Blob Data Contributor) but do not have — and by design do not want — ACL-management permissions. As a result, these framework-generated setPermission() calls fail with 403 AuthorizationPermissionMismatch, breaking otherwise well-configured workloads.
There is currently no way to disable this driver-side behavior without also disabling the explicit ACL management APIs, which some customers still rely on.
Proposed change :-
Introduce an opt-in ABFS configuration:
When enabled on an HNS-enabled account, AzureBlobFileSystem#setPermission() is treated as a pure no-op: the call returns successfully without contacting the backend.fs.azure.rbac.only
true
Disabled by default — no behavior change for existing deployments.
HNS-only — on non-HNS accounts the flag has no effect; existing driver semantics are preserved.
Scoped strictly to setPermission() — no other API is gated by this flag.
setPermission()returns success as a no-op that touches nothing. All other operations still fail with 403.Storage Blob Data ReadersetPermission(), and ACL APIs fail with 403.setPermission()returns success as a no-op. Reader still cannot read anything they couldn't already read.Storage Blob Data ContributorsetPermission()and ACL APIs fail with 403.setPermission()succeeds as a no-op. ACL APIs still fail with 403. This is the target use case.Storage Blob Data Owner(or ACL-management rights)setPermission(), succeed and are applied.setPermission()is a no-op (does not apply). ACL APIs,setOwner(), reads, and writes all continue to work normally.