Add feature flags to disable discovery, findings, and resource watch (#2020)

* Add env vars to disable findings persistence, discovery, and resource watch writes

DISABLE_FINDINGS_PERSISTENCE - skip writing findings to Issues/Evidence tables
DISABLE_DISCOVERY - skip the periodic resource discovery loop (keeps cluster status)
DISABLE_RESOURCE_WATCH_PERSISTENCE - skip real-time K8s watch event DB writes

https://claude.ai/code/session_015jBvu6MGAga8iFmnWmiA2g

* handle no resource on pod investigator

---------

Co-authored-by: Claude <noreply@anthropic.com>

Author: arikalon1

playbooks/robusta_playbooks/pod_investigator_enricher.py

@@ -41,6 +41,10 @@ def pod_issue_investigator(event: KubernetesResourceEvent):
         The only supported resources for investigation are "Deployment", "DaemonSet", "ReplicaSet", "Pod", "StatefulSet", "Job"
     """
     resource = event.get_resource()
+    if not resource:
+        logging.warning("Cannot run pod issue investigator with no resource")
+        return
+
     if resource.kind not in supported_resources:
         raise ActionException(
             ErrorCodes.RESOURCE_NOT_SUPPORTED, f"Pod investigator is not supported for resource {resource.kind}"

src/robusta/core/model/env_vars.py

@@ -99,6 +99,9 @@ def load_bool(env_var, default: bool):
 DISCOVERY_POD_OWNED_PODS = load_bool("DISCOVERY_POD_OWNED_PODS", False)
 
 DISABLE_HELM_MONITORING = load_bool("DISABLE_HELM_MONITORING", False)
+DISABLE_FINDINGS_PERSISTENCE = load_bool("DISABLE_FINDINGS_PERSISTENCE", False)
+DISABLE_DISCOVERY = load_bool("DISABLE_DISCOVERY", False)
+DISABLE_RESOURCE_WATCH_PERSISTENCE = load_bool("DISABLE_RESOURCE_WATCH_PERSISTENCE", False)
 
 PROMETHEUS_ERROR_LOG_PERIOD_SEC = int(os.environ.get("DISCOVERY_MAX_BATCHES", 14400))
 

src/robusta/core/sinks/robusta/robusta_sink.py

@@ -17,6 +17,9 @@
 from robusta.core.model.cluster_status import ActivityStats, ClusterStats, ClusterStatus
 from robusta.core.model.env_vars import (
     CLUSTER_STATUS_PERIOD_SEC,
+    DISABLE_DISCOVERY,
+    DISABLE_FINDINGS_PERSISTENCE,
+    DISABLE_RESOURCE_WATCH_PERSISTENCE,
     DISCOVERY_CHECK_THRESHOLD_SEC,
     DISCOVERY_PERIOD_SEC,
     DISCOVERY_WATCHDOG_CHECK_SEC,
@@ -87,6 +90,13 @@ def __init__(self, sink_config: RobustaSinkConfigWrapper, registry):
             self.signing_key,
         )
 
+        if DISABLE_FINDINGS_PERSISTENCE:
+            logging.info("DISABLE_FINDINGS_PERSISTENCE is set - findings will not be persisted to the platform database")
+        if DISABLE_DISCOVERY:
+            logging.info("DISABLE_DISCOVERY is set - resource discovery will not run")
+        if DISABLE_RESOURCE_WATCH_PERSISTENCE:
+            logging.info("DISABLE_RESOURCE_WATCH_PERSISTENCE is set - real-time resource watch writes are disabled")
+
         self.first_prometheus_alert_time = 0
         self.last_send_time = 0
         self.__discovery_period_sec = DISCOVERY_PERIOD_SEC
@@ -228,6 +238,8 @@ def handle_service_diff(
         new_resource: Union[Deployment, DaemonSet, StatefulSet, ReplicaSet, Pod, Node],
         operation: K8sOperationType,
     ):
+        if DISABLE_RESOURCE_WATCH_PERSISTENCE:
+            return
         try:
             if isinstance(new_resource, (Deployment, DaemonSet, StatefulSet, ReplicaSet, Pod)):
                 self.__publish_single_service(Discovery.create_service_info_from_hikaru(new_resource), operation)
@@ -244,6 +256,8 @@ def handle_service_diff(
             )
 
     def write_finding(self, finding: Finding, platform_enabled: bool):
+        if DISABLE_FINDINGS_PERSISTENCE:
+            return
         self.dal.persist_finding(finding)
 
     def __publish_single_service(self, new_service: ServiceInfo, operation: K8sOperationType):
@@ -626,18 +640,21 @@ def __discovery_watchdog(self):
 
     def __discover_cluster(self):
         logging.info("Cluster discovery initialized")
-        get_history = self.__should_run_history()
+        get_history = self.__should_run_history() if not DISABLE_DISCOVERY else False
         self.last_namespace_discovery = 0
         while self.__active:
             start_t = time.time()
             self.__periodic_cluster_status()
-            discovery_results = self.__discover_resources()
 
-            if get_history:
-                self.__get_events_history()
-                get_history = False
-            if discovery_results and discovery_results.helm_releases:
-                self.__send_helm_release_events(release_data=discovery_results.helm_releases)
+            if not DISABLE_DISCOVERY:
+                discovery_results = self.__discover_resources()
+
+                if get_history:
+                    self.__get_events_history()
+                    get_history = False
+                if discovery_results and discovery_results.helm_releases:
+                    self.__send_helm_release_events(release_data=discovery_results.helm_releases)
+
             duration = round(time.time() - start_t)
             sleep_dur = min(max(self.__discovery_period_sec, 3 * duration), 300)