Try to cleanup mac executor even more

Author: Pijukatel

.github/workflows/run_code_checks.yaml

@@ -40,9 +40,18 @@ jobs:
       HTTPBIN_URL: 'asd'
 
     steps:
-      - name: Disable Spotlight indexing
+      - name: Os cleanup
         if: runner.os == 'macOS'
-        run: sudo mdutil -i off /
+        run: |
+          sudo mdutil -i off /
+          launchctl unload -w /System/Library/LaunchAgents/com.apple.ecosystemanalyticsd.plist
+          sudo killall ecosystemanalyticsd
+          sudo killall Finder
+          sudo killall mds
+          sudo killall mds_stores
+          sudo killall mds_worker
+          sudo killall mdworker
+          sudo killall mdworker_shared
 
       - name: Checkout repository
         uses: actions/checkout@v4

src/crawlee/_utils/system.py

@@ -5,6 +5,7 @@
 from contextlib import suppress
 from datetime import datetime, timezone
 from logging import getLogger
+from subprocess import run
 from typing import Annotated
 
 import psutil
@@ -96,7 +97,10 @@ def get_cpu_info() -> CpuInfo:
 
     cpu_percent = psutil.cpu_percent(percpu=True)
     logger.info(f'Calling get_cpu_info()...: {cpu_percent}')
-    return CpuInfo(used_ratio=sum(cpu_percent)/len(cpu_percent) / 100)
+    ratio = sum(cpu_percent)/len(cpu_percent) / 100
+    if ratio>0.95:
+        print_ps()
+    return CpuInfo(used_ratio=ratio)
 
 
 def get_memory_info() -> MemoryInfo:
@@ -124,3 +128,7 @@ def get_memory_info() -> MemoryInfo:
         current_size=ByteSize(current_size_bytes),
         system_wide_used_size=ByteSize(vm.total - vm.available),
     )
+
+
+def print_ps():
+    run("ps -awxo pid,%cpu,comm", shell=True)

tests/unit/crawlers/_basic/test_basic_crawler.py

@@ -23,6 +23,7 @@
 from crawlee._request import Request, RequestState
 from crawlee._types import BasicCrawlingContext, EnqueueLinksKwargs, HttpMethod
 from crawlee._utils.robots import RobotsTxtFile
+from crawlee._utils.system import print_ps
 from crawlee.configuration import Configuration
 from crawlee.crawlers import BasicCrawler
 from crawlee.errors import RequestCollisionError, SessionError, UserDefinedErrorHandlerError
@@ -1293,11 +1294,7 @@ async def test_timeout_in_handler(sleep_type: str, _) -> None:
     from asyncio import timeout  # type:ignore[attr-defined] # noqa: PLC0415
 
     # Debug CPu usage before starting the test
-    run("ps -ewfaxo comm,user,pid,%cpu,cmd", shell=True)
-
-    run("ps -awxo pid,%cpu,comm", shell=True)
-
-
+    print_ps()
 
     handler_timeout = timedelta(seconds=1)
     max_request_retries = 3