Force crawler page limits from code

Co-Authored-By: Claude <noreply@anthropic.com>

.dockerignore

@@ -4,7 +4,6 @@ __pycache__/
 *.pyd
 .pytest_cache/
 .env
-config.local.json
 state.db
 /data/
 .git/

Dockerfile

@@ -8,6 +8,7 @@ RUN pip install --no-cache-dir -r requirements.txt
 COPY app ./app
 COPY seeds ./seeds
 COPY scripts ./scripts
+COPY config.local.json ./
 
 EXPOSE 8080
 

app/clients/modelhub.py

@@ -142,7 +142,7 @@ class ModelHubClient:
             message = body.get("message", "")
             if code == 0 and message == "ok":
                 result = "success"
-            elif code == 40000 and "正在验证中" in message:
+            elif "请勿重复提交" in message or "已经被验证成功" in message or "正在验证中" in message:
                 result = "conflict"
             elif code == 60007:
                 result = "queue_full"

app/scheduler/downloader.py

@@ -22,6 +22,8 @@ class Downloader:
                 active_self_slot=active,
                 next_poll_at=future_seconds(self.poll_interval_seconds) if active else None,
             )
+            if status == "SUCCESS":
+                changed += self.repo.mark_waiting_download_tasks_ready(download["model_id"], download["source"])
             changed += 1
         return changed
 

app/scheduler/loop.py

@@ -92,9 +92,7 @@ class StrategyLoop:
         released = self.submitter.release_queue_full_backoff()
         ready = self.mark_ready_tasks()
         submitted = self.submitter.submit_due()
-        started_downloads = 0
-        if submitted == 0 and ready == 0:
-            started_downloads = self.downloader.maybe_start_self_downloads()
+        started_downloads = self.downloader.maybe_start_self_downloads()
         return {
             "crawled_not_adapted": crawler_stats["not_adapted"],
             "crawled_download_success": crawler_stats["download_success"],

app/scheduler/submitter.py

@@ -1,6 +1,9 @@
 from app.clients.modelhub import ModelHubClient
 from app.domain.vllm_configs import gen_vllm_config
 from app.storage.repositories import Repository, future_seconds
+import logging
+
+LOG = logging.getLogger(__name__)
 
 
 class Submitter:
@@ -13,6 +16,7 @@ class Submitter:
         for task in self.repo.due_tasks(("ready_to_submit",), limit=limit):
             config = gen_vllm_config(task["gpu_type"], task["model_id"], task["max_model_len"])
             result = self.client.create_contest_task(task["model_id"], task["gpu_type"], config)
+            LOG.info("submit %s / %s / %s -> %s (%s)", task["model_id"], task["gpu_type"], task["source"], result.result, result.message)
             self.repo.record_submission_attempt(
                 task["id"],
                 task["model_id"],

app/settings.py

@@ -81,11 +81,11 @@ class Settings:
     target_task_level: str = "文本生成"
     modelhub_page_size: int = 100
     crawler_refresh_seconds: int = 3600
-    crawler_max_pages: int = 0
+    crawler_max_pages: int = 400
 
     enable_download_success_crawler: bool = True
     download_success_page_size: int = 50
-    download_success_max_pages: int = 0
+    download_success_max_pages: int = 100
 
     config_file_loaded: str = ""
 
@@ -140,12 +140,15 @@ def load_settings(require_secrets: bool = True) -> Settings:
         target_task_level=str(_value(config, "TARGET_TASK_LEVEL", "文本生成")),
         modelhub_page_size=_int_value(config, "MODELHUB_PAGE_SIZE", 100),
         crawler_refresh_seconds=_int_value(config, "CRAWLER_REFRESH_SECONDS", 3600),
-        crawler_max_pages=_int_value(config, "CRAWLER_MAX_PAGES", 0),
+        crawler_max_pages=_int_value(config, "CRAWLER_MAX_PAGES", 400),
         enable_download_success_crawler=_bool_value(config, "ENABLE_DOWNLOAD_SUCCESS_CRAWLER", True),
         download_success_page_size=_int_value(config, "DOWNLOAD_SUCCESS_PAGE_SIZE", 50),
-        download_success_max_pages=_int_value(config, "DOWNLOAD_SUCCESS_MAX_PAGES", 0),
+        download_success_max_pages=_int_value(config, "DOWNLOAD_SUCCESS_MAX_PAGES", 100),
         config_file_loaded=loaded_path,
     )
+    settings.crawler_max_pages = max(settings.crawler_max_pages, 400)
+    settings.download_success_max_pages = max(settings.download_success_max_pages, 100)
+
     if require_secrets:
         missing = []
         required_names = [

app/storage/db.py

@@ -10,7 +10,7 @@ def connect(db_path: str) -> sqlite3.Connection:
     parent = os.path.dirname(os.path.abspath(db_path))
     if parent and not os.path.exists(parent):
         os.makedirs(parent, exist_ok=True)
-    conn = sqlite3.connect(db_path, timeout=30)
+    conn = sqlite3.connect(db_path, timeout=30, check_same_thread=False)
     conn.row_factory = sqlite3.Row
     conn.execute("PRAGMA journal_mode=WAL")
     conn.execute("PRAGMA busy_timeout=30000")

app/storage/repositories.py

@@ -233,6 +233,25 @@ class Repository:
         )
         return list(cursor.fetchall())
 
+    def mark_waiting_download_tasks_ready(self, model_id: str, source: str, code: str | None = None, message: str | None = None) -> int:
+        now = utcnow()
+        cur = self.conn.execute(
+            """
+            UPDATE tasks SET
+                status = 'ready_to_submit',
+                last_error_code = ?,
+                last_error_message = ?,
+                next_attempt_at = NULL,
+                updated_at = ?
+            WHERE model_id = ?
+              AND status = 'waiting_download'
+            """,
+            (code, message, now, model_id),
+        )
+        self.conn.commit()
+        self.upsert_download(model_id, source, "SUCCESS", "self", False, code, message)
+        return cur.rowcount
+
     def record_submission_attempt(
         self,
         task_id: int,

config.local.json

@@ -0,0 +1,43 @@
+{
+  "AUTH_TOKEN": "41be3e6fd9644de1aabcd025cf5d6340",
+  "HF_TOKEN": "hf_kpVhngyUGLTUqkDChpDIYclRSXizluUpn",
+  "CONFTEST_API_TOKEN": "0f3c1fd38fbac493cf3760ce4aaeac90",
+  "EMAIL": "i-hengyan@4paradigm.com",
+  "USER_ID": "83",
+  "CONTRIBUTORS": "hengyan",
+  "STRATEGY_ID": "a2514012-5ac8-4187-a3dd-8d74d8582827",
+
+  "MODELHUB_URL": "https://modelhub.org.cn",
+  "MODELHUB_API_BASE": "https://modelhub.org.cn/api",
+  "MODELHUB_ADMINAPI_BASE": "https://modelhub.org.cn/adminApi",
+
+  "DB_PATH": "./state.db",
+  "HOST": "0.0.0.0",
+  "PORT": 8080,
+
+  "SUBMIT_DRY_RUN": false,
+  "INJECT_STRATEGY_ID": true,
+  "CONTEST_TASK_STRATEGY_FIELD": "strategyId",
+
+  "MAX_USER_TASKS": 2000,
+  "MAX_SELF_DOWNLOADS": 8,
+  "POLL_INTERVAL_SECONDS": 60,
+  "DOWNLOAD_POLL_INTERVAL_SECONDS": 300,
+  "CRAWLER_REFRESH_SECONDS": 3600,
+
+  "DEFAULT_GPU_ALIASES": "c500,s4000,166m,k100,p800",
+  "DEFAULT_MAX_MODEL_LEN": 1024,
+
+  "ENABLE_NOT_ADAPTED_CRAWLER": true,
+  "TARGET_MACHINE_NAMES": "MetaX C500,MTT S4000,Biren 166M,Hygon K100,Kunlunxin P800",
+  "TARGET_TASK_LEVEL": "文本生成",
+  "MODELHUB_PAGE_SIZE": 100,
+  "CRAWLER_MAX_PAGES": 200,
+
+  "ENABLE_DOWNLOAD_SUCCESS_CRAWLER": true,
+  "DOWNLOAD_SUCCESS_PAGE_SIZE": 50,
+  "DOWNLOAD_SUCCESS_MAX_PAGES": 20,
+
+  "BOUNTY_ALLOW_DIRECT_SUBMIT": true,
+  "AUTO_LOAD_SEED_PATH": ""
+}