[P1] Add Memory Links Cleanup #11

New issue

Closed

opened 2025-12-05 13:49:13 +00:00 by blightbow · 1 comment

blightbow commented

2025-12-05 13:49:13 +00:00

Owner

Problem

memory_links dictionary grows indefinitely
Never cleaned when Mem0 entries are deleted
consolidated_entry_ids set only grows, orphaned IDs never pruned

Suggested Fix

Prune orphaned links when Mem0 entries are deleted.

Priority

P1 — High Priority

Source

Architecture Audit 2025-12-03, Section 2: Memory Leak Potential

## Problem - `memory_links` dictionary grows indefinitely - Never cleaned when Mem0 entries are deleted - `consolidated_entry_ids` set only grows, orphaned IDs never pruned ## Suggested Fix Prune orphaned links when Mem0 entries are deleted. ## Priority **P1 — High Priority** ## Source Architecture Audit 2025-12-03, Section 2: Memory Leak Potential

blightbow added the

labels

2025-12-05 13:49:13 +00:00

blightbow commented

2025-12-06 04:38:37 +00:00

Author

Owner

Resolution Summary

Implemented two cleanup mechanisms to prevent unbounded growth of memory links and consolidated entry IDs during the dreaming phase:

1. Orphaned Memory Links Cleanup (`rag_memory.py`)

Added prune_orphaned_memory_links() function that:

Gets all current memory IDs from Mem0 via memory_client.get_all()
Removes memory_links entries whose keys reference deleted memories
Cleans linked_to arrays by removing references to deleted memories
Called during dreaming phase after run_dreaming_tick()

2. Consolidated Entry IDs Cleanup (`helpers.py`)

Enhanced prune_low_importance_entries() to:

Track which journal entries are being pruned
Remove pruned entry IDs from consolidated_entry_ids set
Prevents the consolidated set from growing with stale references

Tests Added

11 new tests in test_memory_link_cleanup.py:

Skip behavior when memory client unavailable
Empty memory_links handling
Orphaned key removal
linked_to array cleanup
Combined cleanup scenarios
Error handling for Mem0 failures
consolidated_entry_ids synchronization

Commit: b412b8156

## Resolution Summary Implemented two cleanup mechanisms to prevent unbounded growth of memory links and consolidated entry IDs during the dreaming phase: ### 1. Orphaned Memory Links Cleanup (`rag_memory.py`) Added `prune_orphaned_memory_links()` function that: - Gets all current memory IDs from Mem0 via `memory_client.get_all()` - Removes `memory_links` entries whose keys reference deleted memories - Cleans `linked_to` arrays by removing references to deleted memories - Called during dreaming phase after `run_dreaming_tick()` ### 2. Consolidated Entry IDs Cleanup (`helpers.py`) Enhanced `prune_low_importance_entries()` to: - Track which journal entries are being pruned - Remove pruned entry IDs from `consolidated_entry_ids` set - Prevents the consolidated set from growing with stale references ### Tests Added 11 new tests in `test_memory_link_cleanup.py`: - Skip behavior when memory client unavailable - Empty memory_links handling - Orphaned key removal - linked_to array cleanup - Combined cleanup scenarios - Error handling for Mem0 failures - consolidated_entry_ids synchronization **Commit:** b412b8156

blightbow closed this issue

2025-12-06 04:38:45 +00:00

blightbow referenced this issue from a commit

2025-12-08 07:35:37 +00:00

Add memory link cleanup to prevent unbounded growth