Schedule periodic attempts to restart collective

If collective is lost due to connectivity, this will cause
occasional attempts to bring it back.

confluent_server/confluent/collective/manager.py

@@ -35,6 +35,7 @@ except ImportError:
 currentleader = None
 cfginitlock = None
 follower = None
+retrythread = None
 
 class ContextBool(object):
     def __init__(self):
@@ -168,6 +169,7 @@ def get_myname():
 
 def handle_connection(connection, cert, request, local=False):
     global currentleader
+    global retrythread
     operation = request['operation']
     if cert:
         cert = crypto.dump_certificate(crypto.FILETYPE_ASN1, cert)
@@ -340,6 +342,9 @@ def handle_connection(connection, cert, request, local=False):
                              connection.getpeername()[0])
             connection.close()
             return
+        if retrythread:
+            retrythread.cancel()
+            retrythread = None
         cfm.update_collective_address(request['name'],
                                       connection.getpeername()[0])
         tlvdata.send(connection, cfm._dump_keys(None, False))
@@ -409,6 +414,7 @@ def startup():
 
 def start_collective():
     global follower
+    global retrythread
     if follower:
         follower.kill()
         follower = None
@@ -422,10 +428,7 @@ def start_collective():
         if connect_to_leader(name=myname, leader=ldrcandidate):
             break
     else:
-        for member in cfm.list_collective():
-            if member == myname:
-                continue
-            eventlet.spawn_n(try_assimilate,
-                             cfm.get_collective_member(member)['address'])
+        retrythread = eventlet.spawn_after(30 + random.random(),
+                                           start_collective)