fix: properly use gevent and allow configuring k8s cache bypass

renku_notebooks/__init__.py

@@ -16,6 +16,10 @@
 # limitations under the License.
 """Notebooks service flask app."""
 
+from gevent import monkey
+
+monkey.patch_all()
+
 import os
 
 from apispec import APISpec

renku_notebooks/api/classes/k8s_client.py

@@ -405,11 +405,13 @@ def __init__(
         renku_ns_client: NamespacedK8sClient,
         username_label: str,
         session_ns_client: Optional[NamespacedK8sClient] = None,
+        bypass_cache_on_failure: bool = True,
     ):
         self.js_cache = js_cache
         self.renku_ns_client = renku_ns_client
         self.username_label = username_label
         self.session_ns_client = session_ns_client
+        self.bypass_cache_on_failure = bypass_cache_on_failure
         if not self.username_label:
             raise ProgrammingError("username_label has to be provided to K8sClient")
 
@@ -421,6 +423,8 @@ def list_servers(self, safe_username: str) -> list[dict[str, Any]]:
         try:
             return self.js_cache.list_servers(safe_username)
         except JSCacheError:
+            if not self.bypass_cache_on_failure:
+                raise
             logging.warning(f"Skipping the cache to list servers for user: {safe_username}")
             label_selector = f"{self.username_label}={safe_username}"
             return self.renku_ns_client.list_servers(label_selector) + (
@@ -436,6 +440,8 @@ def get_server(self, name: str, safe_username: str) -> Optional[dict[str, Any]]:
         try:
             server = self.js_cache.get_server(name)
         except JSCacheError:
+            if not self.bypass_cache_on_failure:
+                raise
             output = []
             res = None
             if self.session_ns_client is not None:

renku_notebooks/config/__init__.py

@@ -101,6 +101,7 @@ def __post_init__(self):
             renku_ns_client=renku_ns_client,
             session_ns_client=session_ns_client,
             username_label=username_label,
+            bypass_cache_on_failure=self.k8s.bypass_cache_on_failure,
         )
         self._crc_validator = None
         self._storage_validator = None
@@ -160,9 +161,7 @@ def get_config(default_config: str) -> _NotebooksConfig:
     config = dataconf.multi.string(default_config)
     if config_file:
         config = config.file(config_file)
-    notebooks_config: _NotebooksConfig = config.env("NB_", ignore_unexpected=True).on(
-        _NotebooksConfig
-    )
+    notebooks_config: _NotebooksConfig = config.env("NB_", ignore_unexpected=True).on(_NotebooksConfig)
     return notebooks_config
 
 

renku_notebooks/config/dynamic.py

@@ -231,9 +231,11 @@ class _K8sConfig:
     renku_namespace: str
     sessions_namespace: Optional[str] = None
     enabled: Union[str, bool] = True
+    bypass_cache_on_failure: Union[str, bool] = True
 
     def __post_init__(self):
         self.enabled = _parse_str_as_bool(self.enabled)
+        self.bypass_cache_on_failure = _parse_str_as_bool(self.bypass_cache_on_failure)
 
 
 @dataclass