solve race condition between job and pod creation by adding retries (#…

…1446)

* solve race condition between job and pod creation by adding retries

* pull default config from env vars

src/robusta/core/model/env_vars.py

@@ -115,7 +115,10 @@ def load_bool(env_var, default: bool):
 # lowered case k8s kinds in a json array string. "[\"configmap\", \"secret\"]"
 RESOURCE_YAML_BLOCK_LIST = json.loads(os.environ.get("RESOURCE_YAML_BLOCK_LIST", "[]"))
 
-NAMESPACE_DATA_TTL = int(os.environ.get("NAMESPACE_DATA_TTL", 30*60))  # in seconds
+NAMESPACE_DATA_TTL = int(os.environ.get("NAMESPACE_DATA_TTL", 30 * 60))  # in seconds
 
-PROCESSED_ALERTS_CACHE_TTL = int(os.environ.get("PROCESSED_ALERT_CACHE_TTL", 2*3600))
+PROCESSED_ALERTS_CACHE_TTL = int(os.environ.get("PROCESSED_ALERT_CACHE_TTL", 2 * 3600))
 PROCESSED_ALERTS_CACHE_MAX_SIZE = int(os.environ.get("PROCESSED_ALERTS_CACHE_MAX_SIZE", 100_000))
+
+POD_WAIT_RETRIES = int(os.environ.get("POD_WAIT_RETRIES", 10))
+POD_WAIT_RETRIES_SECONDS = int(os.environ.get("POD_WAIT_RETRIES_SECONDS", 5))

src/robusta/core/sinks/robusta/dal/supabase_dal.py

@@ -566,7 +566,7 @@ def publish_cluster_nodes(self, node_count: int, pod_count: int):
         try:
             self.client.rpc(UPDATE_CLUSTER_NODE_COUNT, data).execute()
         except Exception as e:
-            logging.error(f"Failed to publish node count {data} error: {e}")
+            logging.exception(f"Failed to publish node count {data} error: {e}")
 
         logging.debug(f"cluster nodes: {UPDATE_CLUSTER_NODE_COUNT} => {data}")
 

src/robusta/integrations/kubernetes/custom_models.py

@@ -13,7 +13,13 @@
 from kubernetes.client import ApiException
 from pydantic import BaseModel
 
-from robusta.core.model.env_vars import IMAGE_REGISTRY, INSTALLATION_NAMESPACE, RUNNER_SERVICE_ACCOUNT
+from robusta.core.model.env_vars import (
+    IMAGE_REGISTRY,
+    INSTALLATION_NAMESPACE,
+    POD_WAIT_RETRIES,
+    POD_WAIT_RETRIES_SECONDS,
+    RUNNER_SERVICE_ACCOUNT,
+)
 from robusta.integrations.kubernetes.api_client_utils import (
     SUCCEEDED_STATE,
     exec_shell_command,
@@ -451,13 +457,19 @@ def get_pods(self) -> List[RobustaPod]:
         # we serialize and then deserialize to work around https://github.com/haxsaw/hikaru/issues/15
         return [hikaru.from_dict(pod.to_dict(), cls=RobustaPod) for pod in pods.items]
 
-    def get_single_pod(self) -> RobustaPod:
+    def get_single_pod(self, retries: int = POD_WAIT_RETRIES, wait: int = POD_WAIT_RETRIES_SECONDS) -> RobustaPod:
         """
         like get_pods() but verifies that only one pod is associated with the job and returns that pod
+        if no pods, retry X times with Y seconds wait
         """
         pods = self.get_pods()
+        while retries > 0 and len(pods) == 0:
+            time.sleep(wait)
+            pods = self.get_pods()
+            retries -= 1
+
         if len(pods) != 1:
-            raise Exception(f"got more pods than expected for job {self.metadata.name}: {pods}")
+            raise Exception(f"got {len(pods)} pods. expected 1 for job {self.metadata.name}: {pods}")
         return pods[0]
 
     def create_job_owned_secret(self, job_secret: JobSecret):
@@ -478,7 +490,7 @@ def create_job_owned_secret(self, job_secret: JobSecret):
             metadata=ObjectMeta(name=job_secret.name, ownerReferences=[robusta_owner_reference]), data=job_secret.data
         )
         try:
-            return secret.createNamespacedSecret(job_pod.metadata.namespace).obj
+            secret.createNamespacedSecret(job_pod.metadata.namespace).obj
         except Exception as e:
             logging.error(f"Failed to create secret {job_secret.name}", exc_info=True)
             raise e