Increasing neutron timeout for low memory usage

We are noticing several tests failing in our low memory environment
because of timeout in neutron requests.
As an example the test
tempest.api.compute.servers.test_server_actions.ServerActionsTestJSON
fails because it requests to plug a vif, and send request to neutron,
which responds in more than neutron_url_timeout, and since the option
vif_plugging_is_fatal is set to True as default, the test fails.
Shortly thereafter, checking in neutron log you can see the request,
returning with the proper status, after more than neutron_url_timeout,
however, it's already too late once nova already marked the instance
with error status, and so the test fails.

Closes-Bug: #1641135

Change-Id: If0991c114f199490ac0deb71eb569a42d4711359

diff --git a/environments/low-memory-usage.yaml b/environments/low-memory-usage.yaml
index ad42868..47b2003 100644 (file)
--- a/environments/low-memory-usage.yaml
+++ b/environments/low-memory-usage.yaml
@@ -13,3 +13,6 @@ parameter_defaults:
 
   ApacheMaxRequestWorkers: 32
   ApacheServerLimit: 32
+
+  ControllerExtraConfig:
+      'nova::network::neutron::neutron_url_timeout': '60'