codex-lb 0.1.4__tar.gz → 0.1.5__tar.gz

codex_lb-0.1.5/.github/release-please-manifest.json

@@ -0,0 +1,3 @@
+{
+  ".": "0.1.5"
+}

{codex_lb-0.1.4 → codex_lb-0.1.5}/CHANGELOG.md

@@ -1,5 +1,12 @@
 # Changelog
 
+## [0.1.5](https://github.com/Soju06/codex-lb/compare/v0.1.4...v0.1.5) (2026-01-14)
+
+
+### Bug Fixes
+
+* align rate-limit backoff and reset handling ([4d59650](https://github.com/Soju06/codex-lb/commit/4d596508e5ad13e68aa6e64f9cb32324bd38f07b))
+
 ## [0.1.4](https://github.com/Soju06/codex-lb/compare/v0.1.3...v0.1.4) (2026-01-13)
 
 

{codex_lb-0.1.4 → codex_lb-0.1.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codex-lb
-Version: 0.1.4
+Version: 0.1.5
 Summary: Codex load balancer and proxy for ChatGPT accounts with usage dashboard
 Author-email: Soju06 <qlskssk@gmail.com>
 Maintainer-email: Soju06 <qlskssk@gmail.com>

{codex_lb-0.1.4 → codex_lb-0.1.5}/app/core/balancer/logic.py

@@ -5,7 +5,7 @@ from dataclasses import dataclass
 from typing import Iterable
 
 from app.core.balancer.types import UpstreamError
-from app.core.utils.retry import parse_retry_after
+from app.core.utils.retry import backoff_seconds, parse_retry_after
 from app.db.models import AccountStatus
 
 PERMANENT_FAILURE_CODES = {
@@ -22,7 +22,7 @@ class AccountState:
     account_id: str
     status: AccountStatus
     used_percent: float | None = None
-    reset_at: int | None = None
+    reset_at: float | None = None
     last_error_at: float | None = None
     last_selected_at: float | None = None
     error_count: int = 0
@@ -97,18 +97,11 @@ def handle_rate_limit(state: AccountState, error: UpstreamError) -> None:
     state.status = AccountStatus.RATE_LIMITED
     state.error_count += 1
     state.last_error_at = time.time()
-
-    reset_at = _extract_reset_at(error)
-    if reset_at is not None:
-        state.reset_at = reset_at
-        return
-
     message = error.get("message")
     delay = parse_retry_after(message) if message else None
-    if delay:
-        state.reset_at = int(time.time() + delay)
-    else:
-        state.reset_at = int(time.time() + 300)
+    if delay is None:
+        delay = backoff_seconds(state.error_count)
+    state.reset_at = time.time() + delay
 
 
 def handle_quota_exceeded(state: AccountState, error: UpstreamError) -> None:

codex_lb-0.1.5/app/core/utils/retry.py

@@ -0,0 +1,30 @@
+from __future__ import annotations
+
+import random
+import re
+
+_RETRY_PATTERN = re.compile(r"(?i)try again in\s*(\d+(?:\.\d+)?)\s*(s|ms|seconds?)")
+_BACKOFF_INITIAL_DELAY_MS = 200
+_BACKOFF_FACTOR = 2.0
+_BACKOFF_JITTER_MIN = 0.9
+_BACKOFF_JITTER_MAX = 1.1
+
+
+def parse_retry_after(message: str) -> float | None:
+    match = _RETRY_PATTERN.search(message or "")
+    if not match:
+        return None
+    value = float(match.group(1))
+    unit = match.group(2).lower()
+    if unit == "ms":
+        return value / 1000
+    return value
+
+
+def backoff_seconds(attempt: int) -> float:
+    if attempt < 1:
+        attempt = 1
+    exponent = _BACKOFF_FACTOR ** (attempt - 1)
+    base_ms = _BACKOFF_INITIAL_DELAY_MS * exponent
+    jitter = random.uniform(_BACKOFF_JITTER_MIN, _BACKOFF_JITTER_MAX)
+    return (base_ms * jitter) / 1000.0

{codex_lb-0.1.4 → codex_lb-0.1.5}/app/modules/proxy/load_balancer.py

@@ -20,7 +20,7 @@ from app.modules.usage.repository import UsageRepository
 
 @dataclass
 class RuntimeState:
-    reset_at: int | None = None
+    reset_at: float | None = None
     last_error_at: float | None = None
     last_selected_at: float | None = None
     error_count: int = 0
@@ -179,10 +179,10 @@ def _apply_secondary_quota(
     *,
     status: AccountStatus,
     primary_used: float | None,
-    runtime_reset: int | None,
+    runtime_reset: float | None,
     secondary_used: float | None,
     secondary_reset: int | None,
-) -> tuple[AccountStatus, float | None, int | None]:
+) -> tuple[AccountStatus, float | None, float | None]:
     used_percent = primary_used
     reset_at = runtime_reset
 

{codex_lb-0.1.4 → codex_lb-0.1.5}/app/static/index.js

@@ -569,16 +569,9 @@
 			return acc;
 		}, {});
 
-	const mergeUsageIntoAccounts = (
-		accounts,
-		primaryUsage,
-		secondaryUsage,
-		summary,
-	) => {
+	const mergeUsageIntoAccounts = (accounts, primaryUsage, secondaryUsage) => {
 		const primaryMap = buildUsageIndex(primaryUsage || []);
 		const secondaryMap = buildUsageIndex(secondaryUsage || []);
-		const resetAtPrimary = summary?.primaryWindow?.resetAt ?? null;
-		const resetAtSecondary = summary?.secondaryWindow?.resetAt ?? null;
 		return accounts.map((account) => {
 			const primaryRow = primaryMap[account.id];
 			const secondaryRow = secondaryMap[account.id];
@@ -598,8 +591,8 @@
 						account.usage?.secondaryRemainingPercent ??
 						0,
 				},
-				resetAtPrimary: resetAtPrimary ?? account.resetAtPrimary ?? null,
-				resetAtSecondary: resetAtSecondary ?? account.resetAtSecondary ?? null,
+				resetAtPrimary: account.resetAtPrimary ?? null,
+				resetAtSecondary: account.resetAtSecondary ?? null,
 			};
 		});
 	};
@@ -1191,7 +1184,6 @@
 						accountsResult.value,
 						primaryUsage,
 						secondaryUsage,
-						summary,
 					);
 					this.applyData(
 						{

{codex_lb-0.1.4 → codex_lb-0.1.5}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "codex-lb"
-version = "0.1.4"
+version = "0.1.5"
 description = "Codex load balancer and proxy for ChatGPT accounts with usage dashboard"
 readme = "README.md"
 license = { file = "LICENSE" }

{codex_lb-0.1.4 → codex_lb-0.1.5}/tests/integration/test_accounts_api_extended.py

@@ -2,10 +2,17 @@ from __future__ import annotations
 
 import base64
 import json
+from datetime import datetime, timezone
 
 import pytest
 
 from app.core.auth import fallback_account_id
+from app.core.crypto import TokenEncryptor
+from app.core.utils.time import utcnow
+from app.db.models import Account, AccountStatus
+from app.db.session import SessionLocal
+from app.modules.accounts.repository import AccountsRepository
+from app.modules.usage.repository import UsageRepository
 
 pytestmark = pytest.mark.integration
 
@@ -33,6 +40,32 @@ def _make_auth_json(account_id: str | None, email: str, plan_type: str = "plus")
     return {"tokens": tokens}
 
 
+def _make_account(account_id: str, email: str, plan_type: str = "plus") -> Account:
+    encryptor = TokenEncryptor()
+    return Account(
+        id=account_id,
+        email=email,
+        plan_type=plan_type,
+        access_token_encrypted=encryptor.encrypt("access"),
+        refresh_token_encrypted=encryptor.encrypt("refresh"),
+        id_token_encrypted=encryptor.encrypt("id"),
+        last_refresh=utcnow(),
+        status=AccountStatus.ACTIVE,
+        deactivation_reason=None,
+    )
+
+
+def _iso_utc(epoch_seconds: int) -> str:
+    return (
+        datetime.fromtimestamp(epoch_seconds, tz=timezone.utc)
+        .isoformat()
+        .replace(
+            "+00:00",
+            "Z",
+        )
+    )
+
+
 @pytest.mark.asyncio
 async def test_import_invalid_json_returns_400(async_client):
     files = {"auth_json": ("auth.json", "not-json", "application/json")}
@@ -78,3 +111,53 @@ async def test_delete_account_removes_from_list(async_client):
     assert accounts.status_code == 200
     data = accounts.json()["accounts"]
     assert all(account["accountId"] != "acc_delete" for account in data)
+
+
+@pytest.mark.asyncio
+async def test_accounts_list_includes_per_account_reset_times(async_client, db_setup):
+    primary_a = 1735689600
+    primary_b = 1735693200
+    secondary_a = 1736294400
+    secondary_b = 1736380800
+
+    async with SessionLocal() as session:
+        accounts_repo = AccountsRepository(session)
+        usage_repo = UsageRepository(session)
+
+        await accounts_repo.upsert(_make_account("acc_reset_a", "a@example.com"))
+        await accounts_repo.upsert(_make_account("acc_reset_b", "b@example.com"))
+
+        await usage_repo.add_entry(
+            "acc_reset_a",
+            10.0,
+            window="primary",
+            reset_at=primary_a,
+        )
+        await usage_repo.add_entry(
+            "acc_reset_b",
+            20.0,
+            window="primary",
+            reset_at=primary_b,
+        )
+        await usage_repo.add_entry(
+            "acc_reset_a",
+            30.0,
+            window="secondary",
+            reset_at=secondary_a,
+        )
+        await usage_repo.add_entry(
+            "acc_reset_b",
+            40.0,
+            window="secondary",
+            reset_at=secondary_b,
+        )
+
+    response = await async_client.get("/api/accounts")
+    assert response.status_code == 200
+    payload = response.json()
+    accounts = {item["accountId"]: item for item in payload["accounts"]}
+
+    assert accounts["acc_reset_a"]["resetAtPrimary"] == _iso_utc(primary_a)
+    assert accounts["acc_reset_b"]["resetAtPrimary"] == _iso_utc(primary_b)
+    assert accounts["acc_reset_a"]["resetAtSecondary"] == _iso_utc(secondary_a)
+    assert accounts["acc_reset_b"]["resetAtSecondary"] == _iso_utc(secondary_b)

{codex_lb-0.1.4 → codex_lb-0.1.5}/tests/unit/test_load_balancer.py

@@ -38,19 +38,23 @@ def test_select_account_skips_rate_limited_until_reset():
 
 
 def test_handle_rate_limit_sets_reset_at_from_message():
+    now = time.time()
     state = AccountState("a", AccountStatus.ACTIVE, used_percent=5.0)
     handle_rate_limit(state, {"message": "Try again in 1.5s"})
     assert state.status == AccountStatus.RATE_LIMITED
     assert state.reset_at is not None
+    delay = state.reset_at - now
+    assert 1.2 <= delay <= 2.0
 
 
-def test_handle_rate_limit_uses_resets_in_seconds():
+def test_handle_rate_limit_uses_backoff_when_no_delay():
     now = time.time()
     state = AccountState("a", AccountStatus.ACTIVE, used_percent=5.0)
-    handle_rate_limit(state, {"resets_in_seconds": 10})
+    handle_rate_limit(state, {"message": "Rate limit exceeded."})
     assert state.status == AccountStatus.RATE_LIMITED
     assert state.reset_at is not None
-    assert int(now) <= state.reset_at <= int(now) + 15
+    delay = state.reset_at - now
+    assert 0.15 <= delay <= 0.3
 
 
 def test_handle_quota_exceeded_sets_used_percent():

codex_lb-0.1.4/.github/release-please-manifest.json

@@ -1,3 +0,0 @@
-{
-  ".": "0.1.4"
-}

codex_lb-0.1.4/app/core/utils/retry.py

@@ -1,16 +0,0 @@
-from __future__ import annotations
-
-import re
-
-_RETRY_PATTERN = re.compile(r"(?i)try again in\s*(\d+(?:\.\d+)?)\s*(s|ms|seconds?)")
-
-
-def parse_retry_after(message: str) -> float | None:
-    match = _RETRY_PATTERN.search(message or "")
-    if not match:
-        return None
-    value = float(match.group(1))
-    unit = match.group(2).lower()
-    if unit == "ms":
-        return value / 1000
-    return value