just warn if Heapster is not deployed (Minikube is missing it)

kube_ops_view/cluster_discovery.py

@@ -1,11 +1,11 @@
+import logging
+import re
 import time
 from pathlib import Path
 from urllib.parse import urljoin
 
 import kubernetes.client
 import kubernetes.config
-import logging
-import re
 import requests
 import tokens
 from requests.auth import AuthBase

kube_ops_view/kubernetes.py

@@ -4,6 +4,8 @@ from urllib.parse import urljoin
 
 import requests
 
+from .utils import get_short_error_message
+
 logger = logging.getLogger(__name__)
 
 session = requests.Session()
@@ -90,8 +92,8 @@ def query_kubernetes_cluster(cluster):
         else:
             for metrics in data['items']:
                 nodes[metrics['metadata']['name']]['usage'] = metrics['usage']
-    except:
+    except Exception as e:
-        logger.exception('Failed to get node metrics')
+        logger.warning('Failed to query node metrics {}: {}'.format(cluster.id, get_short_error_message(e)))
     try:
         response = request(cluster, '/api/v1/namespaces/kube-system/services/heapster/proxy/apis/metrics/v1alpha1/pods')
         response.raise_for_status()
@@ -106,6 +108,6 @@ def query_kubernetes_cluster(cluster):
                         for container_metrics in metrics['containers']:
                             if container['name'] == container_metrics['name']:
                                 container['resources']['usage'] = container_metrics['usage']
-    except:
+    except Exception as e:
-        logger.exception('Failed to get pod metrics')
+        logger.warning('Failed to query pod metrics for cluster {}: {}'.format(cluster.id, get_short_error_message(e)))
     return {'id': cluster_id, 'api_server_url': api_server_url, 'nodes': nodes, 'unassigned_pods': unassigned_pods}

kube_ops_view/update.py

@@ -6,6 +6,7 @@ import json_delta
 import requests.exceptions
 
 from .backoff import expo, random_jitter
+from .utils import get_short_error_message
 
 logger = logging.getLogger(__name__)
 
@@ -14,19 +15,6 @@ def calculate_backoff(tries: int):
     return random_jitter(expo(tries, factor=2, max_value=60), jitter=4)
 
 
-def get_short_error_message(e: requests.exceptions.RequestException):
-    '''Generate a reasonable short message why the HTTP request failed'''
-
-    if e.response is not None:
-        # e.g. "401 Unauthorized"
-        return '{} {}'.format(e.response.status_code, e.response.reason)
-    elif isinstance(e, requests.exceptions.ConnectionError):
-        # e.g. "ConnectionError" or "ConnectTimeout"
-        return e.__class__.__name__
-    else:
-        return str(e)
-
-
 def handle_query_failure(e: Exception, cluster, backoff: dict):
     if not backoff:
         backoff = {}
@@ -34,11 +22,10 @@ def handle_query_failure(e: Exception, cluster, backoff: dict):
     backoff['tries'] = tries
     wait_seconds = calculate_backoff(tries)
     backoff['next_try'] = time.time() + wait_seconds
+    message = get_short_error_message(e)
     if isinstance(e, requests.exceptions.RequestException):
-        message = get_short_error_message(e)
         log = logger.error
     else:
-        message = str(e)
         log = logger.exception
     log('Failed to query cluster {} ({}): {} (try {}, wait {} seconds)'.format(
         cluster.id, cluster.api_server_url, message, tries, round(wait_seconds)))

kube_ops_view/utils.py

@@ -0,0 +1,14 @@
+import requests.exceptions
+
+
+def get_short_error_message(e: Exception):
+    '''Generate a reasonable short message why the HTTP request failed'''
+
+    if isinstance(e, requests.exceptions.RequestException) and e.response is not None:
+        # e.g. "401 Unauthorized"
+        return '{} {}'.format(e.response.status_code, e.response.reason)
+    elif isinstance(e, requests.exceptions.ConnectionError):
+        # e.g. "ConnectionError" or "ConnectTimeout"
+        return e.__class__.__name__
+    else:
+        return str(e)