Skip to content

Finish the Life Cycle of Large Binaries #4123

@kunwp1

Description

@kunwp1

Feature Summary

clearExecutionResources unconditionally calls LargeBinaryManager.deleteAllObjects(), which deletes all large-binary objects in the shared texera-large-binaries bucket even though the method is intended to clean up resources for a single execution eid. This means any user action that triggers cleanup for one workflow execution can erase all other users’ large-binary data, violating tenant isolation and causing global data loss. Instead of a bucket-wide wipe, track the specific LargeBinary URIs associated with the given eid and delete only those objects; reserve any global deleteAllObjects call for tightly controlled, admin-only maintenance flows, if needed at all.

Proposed Solution or Design

In the future, we plan to introduce another abstraction layer that hides the LargeBinary API details from users and lets the system manage large binaries automatically. There, the system should be able to keep track of all the large binaries for each execution and delete them that are scoped to the specific executions.

Impact / Priority

(P2)Medium – useful enhancement

Affected Area

Storage / Metadata

Metadata

Metadata

Assignees

Labels

No labels
No labels

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions