[ARVADOS] created: 1decc19f47cacc6b0fe07daa2011fba565314cc9

Git user git at public.curoverse.com
Thu Mar 23 13:30:48 EDT 2017


        at  1decc19f47cacc6b0fe07daa2011fba565314cc9 (commit)


commit 1decc19f47cacc6b0fe07daa2011fba565314cc9
Author: Peter Amstutz <peter.amstutz at curoverse.com>
Date:   Thu Mar 23 13:30:35 2017 -0400

    11323: Don't try to offer_arvados_pair on unpaired nodes which are being shut down.

diff --git a/services/nodemanager/arvnodeman/daemon.py b/services/nodemanager/arvnodeman/daemon.py
index b4f1784..c0c4945 100644
--- a/services/nodemanager/arvnodeman/daemon.py
+++ b/services/nodemanager/arvnodeman/daemon.py
@@ -232,7 +232,7 @@ class NodeManagerDaemonActor(actor_class):
     def try_pairing(self):
         for record in self.cloud_nodes.unpaired():
             for arv_rec in self.arvados_nodes.unpaired():
-                if record.actor.offer_arvados_pair(arv_rec.arvados_node).get():
+                if record.actor and record.actor.offer_arvados_pair(arv_rec.arvados_node).get():
                     self._pair_nodes(record, arv_rec.arvados_node)
                     break
 

commit 6280916d9db0ca526620bdb7a7197ca6131534f8
Author: Peter Amstutz <peter.amstutz at curoverse.com>
Date:   Thu Mar 23 13:23:54 2017 -0400

    11325: Remove "broken node" check.  Assume if the node really isn't
    functioning, it should be "down" in SLURM anyway.

diff --git a/services/nodemanager/arvnodeman/computenode/dispatch/__init__.py b/services/nodemanager/arvnodeman/computenode/dispatch/__init__.py
index fc3ff05..71f9083 100644
--- a/services/nodemanager/arvnodeman/computenode/dispatch/__init__.py
+++ b/services/nodemanager/arvnodeman/computenode/dispatch/__init__.py
@@ -348,10 +348,6 @@ class ComputeNodeMonitorActor(config.actor_class):
         if self.arvados_node is None:
             return 'unpaired'
 
-        # This node is indicated as non-functioning by the cloud
-        if self._cloud.broken(self.cloud_node):
-            return 'down'
-
         state = self.arvados_node['crunch_worker_state']
 
         # If state information is not available because it is missing or the

-----------------------------------------------------------------------


hooks/post-receive
-- 




More information about the arvados-commits mailing list