cleveragents/cleveragents-core

Fork 3

BUG-HUNT: [resource] CostTracker._daily_costs dict grows unboundedly leaking memory across long-running sessions #7374

New issue

Open

opened 2026-04-10 18:28:09 +00:00 by HAL9000 · 3 comments

HAL9000 commented

2026-04-10 18:28:09 +00:00

Owner

Bug Report: [resource] CostTracker._daily_costs dict accumulates indefinitely, growing unbounded over long-running server sessions

Severity Assessment

Impact: The _daily_costs dictionary in CostTracker stores one entry per day with the key being date.today().isoformat(). In a long-running server process (weeks/months), this dict grows indefinitely, one entry per day. While individual entries are small (one float per day), this is a resource leak that has no cleanup mechanism.
Likelihood: High — any long-running deployment will accumulate entries. After a year, the dict has 365 entries. After 10 years, 3650 entries.
Priority: Medium

Location

File: src/cleveragents/providers/cost_tracker.py
Function/Class: CostTracker.record_usage()
Lines: ~130-145

Description

The CostTracker uses a dictionary _daily_costs: dict[str, float] to track per-day spending:

with self._daily_costs_lock:
    today_key = date.today().isoformat()
    self._daily_costs[today_key] = self._daily_costs.get(today_key, 0.0) + cost

This dictionary grows one entry per day that any cost is recorded. Since the dict is never pruned, it accumulates indefinitely. While the memory impact is small per day, there are additional problems:

Memory leak: Historical daily costs are never cleaned up. A process running for a year has 365 entries, for 5 years: 1825 entries.
Historical data irrelevance: The only method that reads _daily_costs is check_daily_budget(), which only checks today's spending. Historical entries are never used for anything:

def check_daily_budget(self) -> BudgetCheckResult:
    with self._daily_costs_lock:
        today_key = date.today().isoformat()
        used = self._daily_costs.get(today_key, 0.0)  # Only reads today's entry!

No persistence: Since _daily_costs is in-memory only, it's lost on restart. So historical data serves no purpose even if it were used.

Evidence

class CostTracker:
    def __init__(self, ...):
        self._daily_costs: dict[str, float] = {}  # Never cleared!
    
    def record_usage(self, ...) -> BudgetCheckResult:
        # Accumulates forever:
        with self._daily_costs_lock:
            today_key = date.today().isoformat()
            self._daily_costs[today_key] = self._daily_costs.get(today_key, 0.0) + cost
    
    def check_daily_budget(self) -> BudgetCheckResult:
        with self._daily_costs_lock:
            today_key = date.today().isoformat()
            used = self._daily_costs.get(today_key, 0.0)  # Only uses today's entry
        # All other entries are unused!

After a year with daily usage, _daily_costs would have 365 keys like {"2025-01-01": 1.23, "2025-01-02": 0.45, ..., "2025-12-31": 2.10}. All entries except today's are dead data.

Expected Behavior

The dictionary should only retain the current day's entry (or at most the last N days if historical comparison is needed). A cleanup routine should prune old entries:

def record_usage(self, ...) -> BudgetCheckResult:
    cost = self.estimate_cost(...)
    with self._daily_costs_lock:
        today_key = date.today().isoformat()
        self._daily_costs[today_key] = self._daily_costs.get(today_key, 0.0) + cost
        # Keep only today's entry (or last N days)
        self._daily_costs = {today_key: self._daily_costs[today_key]}

Actual Behavior

The _daily_costs dict grows by one entry per day with no cleanup. Historical entries are never used for any purpose.

TDD Note

After this bug issue is verified, a corresponding Type/Testing issue will be created for TDD. The test will use tags: @tdd_issue, @tdd_issue_, and @tdd_expected_fail to prove the bug exists before fixing it.

Automated by CleverAgents Bot
Supervisor: Bug Detection Pool | Agent: bug-hunt-pool-supervisor

## Bug Report: [resource] CostTracker._daily_costs dict accumulates indefinitely, growing unbounded over long-running server sessions ### Severity Assessment - **Impact**: The `_daily_costs` dictionary in `CostTracker` stores one entry per day with the key being `date.today().isoformat()`. In a long-running server process (weeks/months), this dict grows indefinitely, one entry per day. While individual entries are small (one float per day), this is a resource leak that has no cleanup mechanism. - **Likelihood**: High — any long-running deployment will accumulate entries. After a year, the dict has 365 entries. After 10 years, 3650 entries. - **Priority**: Medium ### Location - **File**: `src/cleveragents/providers/cost_tracker.py` - **Function/Class**: `CostTracker.record_usage()` - **Lines**: ~130-145 ### Description The `CostTracker` uses a dictionary `_daily_costs: dict[str, float]` to track per-day spending: ```python with self._daily_costs_lock: today_key = date.today().isoformat() self._daily_costs[today_key] = self._daily_costs.get(today_key, 0.0) + cost ``` This dictionary grows one entry per day that any cost is recorded. Since the dict is never pruned, it accumulates indefinitely. While the memory impact is small per day, there are additional problems: 1. **Memory leak**: Historical daily costs are never cleaned up. A process running for a year has 365 entries, for 5 years: 1825 entries. 2. **Historical data irrelevance**: The only method that reads `_daily_costs` is `check_daily_budget()`, which only checks **today's** spending. Historical entries are never used for anything: ```python def check_daily_budget(self) -> BudgetCheckResult: with self._daily_costs_lock: today_key = date.today().isoformat() used = self._daily_costs.get(today_key, 0.0) # Only reads today's entry! ``` 3. **No persistence**: Since `_daily_costs` is in-memory only, it's lost on restart. So historical data serves no purpose even if it were used. ### Evidence ```python class CostTracker: def __init__(self, ...): self._daily_costs: dict[str, float] = {} # Never cleared! def record_usage(self, ...) -> BudgetCheckResult: # Accumulates forever: with self._daily_costs_lock: today_key = date.today().isoformat() self._daily_costs[today_key] = self._daily_costs.get(today_key, 0.0) + cost def check_daily_budget(self) -> BudgetCheckResult: with self._daily_costs_lock: today_key = date.today().isoformat() used = self._daily_costs.get(today_key, 0.0) # Only uses today's entry # All other entries are unused! ``` After a year with daily usage, `_daily_costs` would have 365 keys like `{"2025-01-01": 1.23, "2025-01-02": 0.45, ..., "2025-12-31": 2.10}`. All entries except today's are dead data. ### Expected Behavior The dictionary should only retain the current day's entry (or at most the last N days if historical comparison is needed). A cleanup routine should prune old entries: ```python def record_usage(self, ...) -> BudgetCheckResult: cost = self.estimate_cost(...) with self._daily_costs_lock: today_key = date.today().isoformat() self._daily_costs[today_key] = self._daily_costs.get(today_key, 0.0) + cost # Keep only today's entry (or last N days) self._daily_costs = {today_key: self._daily_costs[today_key]} ``` ### Actual Behavior The `_daily_costs` dict grows by one entry per day with no cleanup. Historical entries are never used for any purpose. ### Category resource ### TDD Note After this bug issue is verified, a corresponding Type/Testing issue will be created for TDD. The test will use tags: @tdd_issue, @tdd_issue_<this-issue-number>, and @tdd_expected_fail to prove the bug exists before fixing it. --- **Automated by CleverAgents Bot** Supervisor: Bug Detection Pool | Agent: bug-hunt-pool-supervisor

HAL9000 referenced this issue

2026-04-10 18:30:31 +00:00

[AUTO-BUG-SUP] Bug Hunt Status (Cycle 2) #7366

HAL9000 added the

State

Unverified

label

2026-04-10 19:35:14 +00:00

freemo added the

labels

2026-04-12 18:35:38 +00:00

HAL9000 added

and removed

labels

2026-04-14 06:24:09 +00:00

HAL9000 commented

2026-04-14 06:24:09 +00:00

Author

Owner

✅ Verified — Resource leak: CostTracker._daily_costs grows unboundedly. MoSCoW: Should-have. Priority: Medium.

Automated by CleverAgents Bot
Supervisor: Project Owner | Agent: project-owner-pool-supervisor

✅ **Verified** — Resource leak: CostTracker._daily_costs grows unboundedly. MoSCoW: Should-have. Priority: Medium. --- **Automated by CleverAgents Bot** Supervisor: Project Owner | Agent: project-owner-pool-supervisor

HAL9000 commented

2026-04-14 06:29:47 +00:00

Author

Owner

✅ Verified — Resource leak: CostTracker._daily_costs grows unboundedly. MoSCoW: Should-have. Priority: Medium.

Automated by CleverAgents Bot
Supervisor: Project Owner | Agent: project-owner-pool-supervisor