Add endpoint and cron job to aggregate all dimensions and values

appengine/swarming/handlers_backend.py

Index: appengine/swarming/handlers_backend.py
diff --git a/appengine/swarming/handlers_backend.py b/appengine/swarming/handlers_backend.py
index bbcdc6bc90678b32459613a04a3983d496827665..09d7a83b815f31737f8d7f6fc44aede272002b81 100644
--- a/appengine/swarming/handlers_backend.py
+++ b/appengine/swarming/handlers_backend.py
@@ -15,6 +15,7 @@ from google.appengine.api import taskqueue
 import mapreduce_jobs
 from components import decorators
 from components import machine_provider
+from server import bot_management
 from server import config
 from server import lease_management
 from server import stats
@@ -97,6 +98,30 @@ class CronMachineProviderCleanUpHandler(webapp2.RequestHandler):
     lease_management.clean_up_bots()
 
 
+class CronBotDimensionAggregationHandler(webapp2.RequestHandler):
+  """Aggregate all bot dimensions (except id) in the fleet."""

[M-A Ruel, 2016/08/05 14:36:05]

Aggregates
bots

[kjlubick, 2016/08/05 17:38:58]

Done.

+
+  @decorators.require_cronjob
+  def get(self):
+    seen = {}

[M-A Ruel, 2016/08/05 14:36:04]

add:
now = utils.utcnow()

[kjlubick, 2016/08/05 17:38:58]

Done.

+    for b in bot_management.BotInfo.query():
+      for i in b.dimensions_flat:
+        k, v = i.split(':', 1)
+        if k == 'id':

[M-A Ruel, 2016/08/05 14:36:04]

if k != 'id':
  seen.setdefault(k, set()).add(v)

saves you one line. OMG :P

[kjlubick, 2016/08/05 17:38:57]

Done.  The electrons spared rejoiced.

+          continue
+        seen.setdefault(k, set()).add(v)
+    dims = []
+    for k, values in sorted(seen.iteritems()):
+      dims.append(bot_management.DimensionValues(
+        dimension=k,
+        values=sorted(list(values))))
+    logging.info('Saw dimensions %s', dims)
+    aggregate = bot_management.DimensionAggregation(
+      id='current',
+      dimensions=dims)

[M-A Ruel, 2016/08/05 14:36:04]

and save ts=now here so it's at when the query started, not when the entity was
saved, which is an important distinction.

[kjlubick, 2016/08/05 17:38:58]

Done.

+    aggregate.put()
+
+
 class CronMachineProviderPubSubHandler(webapp2.RequestHandler):
   """Listens for Pub/Sub communication from Machine Provider."""
 
@@ -170,6 +195,8 @@ def get_routes():
 
     ('/internal/cron/stats/update', stats.InternalStatsUpdateHandler),
     ('/internal/cron/trigger_cleanup_data', CronTriggerCleanupDataHandler),
+    ('/internal/cron/aggregate_bot_dimensions',
+        CronBotDimensionAggregationHandler),
     ('/internal/cron/machine_provider', CronMachineProviderBotHandler),
     ('/internal/cron/machine_provider_cleanup',
         CronMachineProviderCleanUpHandler),