mirrors/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-04-27 01:11:40 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 5

fix(tools): bound _read_tracker sub-containers + prune _completion_consumed (#11839)

Browse source 
Two accretion-over-time leaks that compound over long CLI / gateway
lifetimes.  Both were flagged in the memory-leak audit.

## file_tools._read_tracker

_read_tracker[task_id] holds three sub-containers that grew unbounded:

  read_history     set of (path, offset, limit) tuples — 1 per unique read
  dedup            dict of (path, offset, limit) → mtime — same growth pattern
  read_timestamps  dict of resolved_path → mtime — 1 per unique path

A CLI session uses one stable task_id for its lifetime, so these were
uncapped.  A 10k-read session accumulated ~1.5MB of tracker state that
the tool no longer needed (only the most recent reads are relevant for
dedup, consecutive-loop detection, and write/patch external-edit
warnings).

Fix: _cap_read_tracker_data() enforces hard caps on each container
after every add.  Defaults: read_history=500, dedup=1000,
read_timestamps=1000.  Eviction is insertion-order (Python 3.7+ dict
guarantee) for the dicts; arbitrary for the set (which only feeds
diagnostic summaries).

## process_registry._completion_consumed

Module-level set that recorded every session_id ever polled / waited /
logged.  No pruning.  Each entry is ~20 bytes, so the absolute leak is
small, but on a gateway processing thousands of background commands
per day the set grows until process exit.

Fix: _prune_if_needed() now discards _completion_consumed entries
alongside the session dict evictions it already performs (both the
TTL-based prune and the LRU-over-cap prune).  Adds a final
belt-and-suspenders pass that drops any dangling entries whose
session_id no longer appears in _running or _finished.

Tests: tests/tools/test_accretion_caps.py — 9 cases
  * Each container bound respected, oldest evicted
  * No-op when under cap (no unnecessary work)
  * Handles missing sub-containers without crashing
  * Live read_file_tool path enforces caps end-to-end
  * _completion_consumed pruned on TTL expiry
  * _completion_consumed pruned on LRU eviction
  * Dangling entries (no backing session) cleared

Broader suite: 3486 tests/tools + tests/cli pass.  The single flake
(test_alias_command_passes_args) reproduces on unchanged main — known
cross-test pollution under suite-order load.
This commit is contained in:
Teknium 2026-04-17 15:53:57 -07:00 • committed by GitHub
parent 0a83187801
commit 3f43aec15d
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 266 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

10
tools/process_registry.py
View file

@ -970,12 +970,22 @@ class ProcessRegistry:
]
for sid in expired:
del self._finished[sid]
self._completion_consumed.discard(sid)
# If still over limit, remove oldest finished
total = len(self._running) + len(self._finished)
if total >= MAX_PROCESSES and self._finished:
oldest_id = min(self._finished, key=lambda sid: self._finished[sid].started_at)
del self._finished[oldest_id]
self._completion_consumed.discard(oldest_id)
# Drop any _completion_consumed entries whose sessions are no longer
# tracked at all — belt-and-suspenders against module-lifetime growth
# on process-registry lookup paths that don't reach the dict prunes.
tracked = self._running.keys() | self._finished.keys()
stale = self._completion_consumed - tracked
if stale:
self._completion_consumed -= stale
# ----- Checkpoint (crash recovery) -----