feat: do not retry for client exceptions

configs/schemas/config.schema.json

@@ -233,6 +233,11 @@
           "title": "Max Retries",
           "type": "integer"
         },
+        "retry_on_client_errors": {
+          "default": true,
+          "title": "Retry On Client Errors",
+          "type": "boolean"
+        },
         "model_max_batch_size": {
           "default": 64,
           "exclusiveMinimum": 0,
@@ -1335,6 +1340,7 @@
         "cache_dir": null,
         "ignore_build_vdb_errors": false,
         "max_retries": 10,
+        "retry_on_client_errors": true,
         "model_max_batch_size": 64,
         "num_threads": 64,
         "pipeline_batch_size": 1024,

src/cve/data_models/config.py

@@ -52,6 +52,7 @@ class GeneralConfig(BaseModel):
     cache_dir: str | None = None
     ignore_build_vdb_errors: bool = False
     max_retries: NonNegativeInt = 10
+    retry_on_client_errors: bool = True
     model_max_batch_size: PositiveInt = 64
     num_threads: PositiveInt = Field(default_factory=os.cpu_count)
     pipeline_batch_size: PositiveInt = 1024

src/cve/pipeline/input.py

@@ -179,7 +179,7 @@ async def _inner():
 
             async with aiohttp.ClientSession() as session:
 
-                intel_retriever = IntelRetriever(session=session)
+                intel_retriever = IntelRetriever(session=session, retry_on_client_errors=run_config.general.retry_on_client_errors)
 
                 intel_coros = [intel_retriever.retrieve(vuln_id=cve.vuln_id) for cve in message.input.scan.vulns]
 
@@ -243,7 +243,7 @@ async def _inner():
 
             async with aiohttp.ClientSession() as session:
 
-                vuln_dep_checker = VulnerableDependencyChecker(session=session, image=image, sbom_list=sbom)
+                vuln_dep_checker = VulnerableDependencyChecker(session=session, image=image, sbom_list=sbom, retry_on_client_errors=run_config.general.retry_on_client_errors)
 
                 await vuln_dep_checker.load_dependencies()
 

src/cve/utils/async_http_utils.py

@@ -31,11 +31,12 @@ async def request_with_retry(session: aiohttp.ClientSession,
                              max_retries: int = 10,
                              sleep_time: float = 0.1,
                              respect_retry_after_header: bool = True,
-                             log_on_error=True) -> typing.AsyncIterator[aiohttp.ClientResponse]:
+                             log_on_error=True,
+                             retry_on_client_errors = True) -> typing.AsyncIterator[aiohttp.ClientResponse]:
     """
     Async version of `morpheus.utils.http_utils.request_with_retry`
     """
-    assert not request_kwargs.get('raise_for_status'), "raise_for_status is cincompatible with `request_with_retry`"
+    assert not request_kwargs.get('raise_for_status'), "raise_for_status is incompatible with `request_with_retry`"
     try_count = 0
     done = False
     while try_count <= max_retries and not done:
@@ -61,6 +62,8 @@ async def request_with_retry(session: aiohttp.ClientSession,
                 actual_sleep_time = max(int(response_headers["Retry-After"]), actual_sleep_time)
             elif respect_retry_after_header and 'X-RateLimit-Reset' in response_headers:
                 actual_sleep_time = max(int(response_headers["X-RateLimit-Reset"]) - time.time(), actual_sleep_time)
+            elif not retry_on_client_errors and response.status < 500:
+                raise e
 
             logger.warning("Error requesting [%d/%d]: (Retry %.1f sec) %s: %s",
                            try_count,

src/cve/utils/clients/first_client.py

@@ -37,13 +37,15 @@ def __init__(self,
                  session: aiohttp.ClientSession | None = None,
                  retry_count: int = 10,
                  sleep_time: float = 0.1,
-                 respect_retry_after_header: bool = True):
+                 respect_retry_after_header: bool = True,
+                 retry_on_client_errors: bool = True):
 
         super().__init__(session=session,
                          base_url=base_url or os.environ.get('FIRST_BASE_URL'),
                          retry_count=retry_count,
                          sleep_time=sleep_time,
-                         respect_retry_after_header=respect_retry_after_header)
+                         respect_retry_after_header=respect_retry_after_header,
+                         retry_on_client_errors=retry_on_client_errors)
 
         self._headers = {'Accept': 'application/json'}
 

src/cve/utils/clients/ghsa_client.py

@@ -41,12 +41,14 @@ def __init__(self,
                  session: aiohttp.ClientSession | None = None,
                  retry_count: int = 10,
                  sleep_time: float = 0.1,
-                 respect_retry_after_header: bool = True):
+                 respect_retry_after_header: bool = True,
+                 retry_on_client_errors: bool = True):
         super().__init__(session=session,
                          base_url=base_url or os.environ.get('GHSA_BASE_URL'),
                          retry_count=retry_count,
                          sleep_time=sleep_time,
-                         respect_retry_after_header=respect_retry_after_header)
+                         respect_retry_after_header=respect_retry_after_header,
+                         retry_on_client_errors=retry_on_client_errors)
 
         self._api_key = api_key or os.environ.get('GHSA_API_KEY', None)
 

src/cve/utils/clients/intel_client.py

@@ -30,7 +30,8 @@ def __init__(self,
                  base_url: str | None = None,
                  retry_count: int = 10,
                  sleep_time: float = 0.1,
-                 respect_retry_after_header: bool = True):
+                 respect_retry_after_header: bool = True,
+                 retry_on_client_errors: bool = True):
 
         if (session is None):
             session = aiohttp.ClientSession()
@@ -44,6 +45,7 @@ def __init__(self,
         self._retry_count = retry_count
         self._sleep_time = sleep_time
         self._respect_retry_after_header = respect_retry_after_header
+        self._retry_on_client_errors = retry_on_client_errors
 
     @classmethod
     @abstractmethod
@@ -80,6 +82,7 @@ async def request(self,
                                       max_retries=self._retry_count,
                                       sleep_time=self._sleep_time,
                                       respect_retry_after_header=self._respect_retry_after_header,
-                                      log_on_error=log_on_error) as response:
+                                      log_on_error=log_on_error,
+                                      retry_on_client_errors=self._retry_on_client_errors) as response:
 
             return await response.json()

src/cve/utils/clients/nvd_client.py

@@ -51,13 +51,15 @@ def __init__(self,
                  session: aiohttp.ClientSession | None = None,
                  retry_count: int = 10,
                  sleep_time: float = 0.1,
-                 respect_retry_after_header: bool = True):
+                 respect_retry_after_header: bool = True,
+                 retry_on_client_errors: bool = True):
 
         super().__init__(session=session,
                          base_url=base_url or os.environ.get('NVD_BASE_URL'),
                          retry_count=retry_count,
                          sleep_time=sleep_time,
-                         respect_retry_after_header=respect_retry_after_header)
+                         respect_retry_after_header=respect_retry_after_header,
+                         retry_on_client_errors=retry_on_client_errors)
 
         self._api_key = api_key or os.environ.get('NVD_API_KEY', None)
 
@@ -87,7 +89,8 @@ async def _get_soup(self, url: str) -> BeautifulSoup:
                                           'url': url,
                                           "skip_auto_headers": {"User-Agent"},
                                       },
-                                      max_retries=self._retry_count) as response:
+                                      max_retries=self._retry_count,
+                                      retry_on_client_errors=self._retry_on_client_errors) as response:
             return BeautifulSoup(await response.text(), 'html.parser')
 
     def _get_cvss_vector_from_metric(self, metrics: dict, metric_version: str) -> str | None:

src/cve/utils/clients/rhsa_client.py

@@ -37,13 +37,15 @@ def __init__(self,
                  session: aiohttp.ClientSession | None = None,
                  retry_count: int = 10,
                  sleep_time: float = 0.1,
-                 respect_retry_after_header: bool = True):
+                 respect_retry_after_header: bool = True,
+                 retry_on_client_errors: bool = True):
 
         super().__init__(session=session,
                          base_url=base_url or os.environ.get('RHSA_BASE_URL'),
                          retry_count=retry_count,
                          sleep_time=sleep_time,
-                         respect_retry_after_header=respect_retry_after_header)
+                         respect_retry_after_header=respect_retry_after_header,
+                         retry_on_client_errors=retry_on_client_errors)
 
     @classmethod
     def default_base_url(cls) -> str:

src/cve/utils/clients/ubuntu_client.py

@@ -35,13 +35,15 @@ def __init__(self,
                  session: aiohttp.ClientSession | None = None,
                  retry_count: int = 10,
                  sleep_time: float = 0.1,
-                 respect_retry_after_header: bool = True):
+                 respect_retry_after_header: bool = True,
+                 retry_on_client_errors: bool = True):
 
         super().__init__(session=session,
                          base_url=base_url or os.environ.get('UBUNTU_BASE_URL'),
                          retry_count=retry_count,
                          sleep_time=sleep_time,
-                         respect_retry_after_header=respect_retry_after_header)
+                         respect_retry_after_header=respect_retry_after_header,
+                         retry_on_client_errors=retry_on_client_errors)
 
     @classmethod
     def default_base_url(cls) -> str:

src/cve/utils/intel_retriever.py

@@ -45,7 +45,8 @@ def __init__(self,
                  nist_api_key: str | None = None,
                  ghsa_api_key: str | None = None,
                  lang_code: str = 'en',
-                 max_retries: int = 10):
+                 max_retries: int = 10,
+                 retry_on_client_errors: bool = True):
         """
         Initialize the NISTCVERetriever with URL templates for vulnerability and CVE details.
         """
@@ -56,13 +57,18 @@ def __init__(self,
         self._nvd_client = NVDClient(api_key=os.environ.get('NVD_API_KEY', nist_api_key),
                                      session=self._session,
                                      lang_code=lang_code,
-                                     retry_count=max_retries)
-        self._first_client = FirstClient(session=self._session, retry_count=max_retries)
+                                     retry_count=max_retries,
+                                     retry_on_client_errors=retry_on_client_errors)
+        self._first_client = FirstClient(session=self._session, retry_count=max_retries,
+                                         retry_on_client_errors=retry_on_client_errors)
         self._ghsa_client = GHSAClient(api_key=os.environ.get('GHSA_API_KEY', ghsa_api_key),
                                        session=self._session,
-                                       retry_count=max_retries)
-        self._rhsa_client = RHSAClient(session=self._session, retry_count=max_retries)
-        self._ubuntu_client = UbuntuClient(session=self._session, retry_count=max_retries)
+                                       retry_count=max_retries,
+                                       retry_on_client_errors=retry_on_client_errors)
+        self._rhsa_client = RHSAClient(session=self._session, retry_count=max_retries,
+                                       retry_on_client_errors=retry_on_client_errors)
+        self._ubuntu_client = UbuntuClient(session=self._session, retry_count=max_retries,
+                                           retry_on_client_errors=retry_on_client_errors)
 
     @asynccontextmanager
     async def _get_session(self, session: aiohttp.ClientSession | None = None):

src/cve/utils/vulnerable_dependency_checker.py

@@ -29,7 +29,6 @@
 
 from ..data_models.cve_intel import CveIntelNvd
 from ..data_models.dependencies import DependencyPackage
-from .async_http_utils import request_with_retry
 from .clients.intel_client import IntelClient
 from .string_utils import package_names_match
 from .url_utils import url_join
@@ -126,13 +125,15 @@ def __init__(self,
                  image: str,
                  sbom_list: list,
                  session: aiohttp.ClientSession | None = None,
-                 respect_retry_after_header: bool = True):
+                 respect_retry_after_header: bool = True,
+                 retry_on_client_errors: bool = True):
 
         super().__init__(
             session=session,
             base_url=base_url or os.environ.get('DEPSDEV_BASE_URL'),
             retry_count=1,  # Service returns 404 if the package is not found. So dont retry
-            respect_retry_after_header=respect_retry_after_header)
+            respect_retry_after_header=respect_retry_after_header,
+            retry_on_client_errors=retry_on_client_errors)
 
         self._semaphore = asyncio.Semaphore(25)