Timestamp metric is no longer explicitly passed by caller for time series panel.

Author: Davide Schiera

examples/dashboard.py

@@ -60,7 +60,6 @@
 panel_name = 'CPU Over Time'
 panel_type = 'timeSeries'
 metrics = [
-  { 'id': 'timestamp' },
   { 'id': 'kubernetes.pod.name' },
   { 'id': 'cpu.used.percent', 'aggregations': { 'time': 'avg', 'group': 'avg' } }
 ]

sdcclient/_client.py

@@ -391,6 +391,17 @@ def add_dashboard_panel(self, dashboard, name, type, metrics, scope=None, sort_b
                                     }
         }
 
+        if type == 'timeSeries':
+            #
+            # In case of a time series, the current dashboard implementation
+            # requires the timestamp to be explicitly specified as "key".
+            # However, this function uses the same abstraction of the data API
+            # that doesn't require to specify a timestamp key (you only need to
+            # specify time window and sampling)
+            #
+            metrics = copy.copy(metrics)
+            metrics.insert(0, { 'id': 'timestamp' })
+
         #
         # Convert list of metrics to format used by Sysdig Cloud
         #