mirror of
https://github.com/hanzoai/tasks.git
synced 2026-11-02 12:22:18 +00:00
## What changed? - Extended tombstone tracking to capture HSM state machine deletions from Operation Log - Added `StateMachinePath` tombstone type for HSM deletions alongside existing activity/timer/etc. tombstones - Maintained existing tombstone batch management and cleanup mechanisms ## Why? Without this: - HSM state machine deletions would not be properly replicated across clusters - Could lead to state inconsistencies between active/passive clusters when using HSM - No way to know if an HSM node was intentionally deleted or is missing due to an error ## Potential risks - Potential memory pressure from large HSM deletion batches - New panic path if operation log access fails Risk Mitigation: - Operation log errors surface quickly via panic rather than silent failure ## Documentation No changes required - this extends an internal implementation detail of how deletions are tracked. ## Is hotfix candidate? No - this should follow normal deployment process.