commit 44b19b9b3b4514c110d41ff72c20e874844fa122
author ghanshyam <ghanshyam.mann@nectechnologies.in> Mon Oct 06 15:59:04 2014 +0900
committer ghanshyam <ghanshyam.mann@nectechnologies.in> Mon Oct 06 15:59:04 2014 +0900
tree 40c91de85417ed7d287b3f9c87d9127f4b322ea3
parent bea1b3c1a223a5a586025bb9e391c77be3c13076

Log AttributeError as 'info' instead of 'exception'

When failure happens in setUpClass, tearDownclass is being called
from base class irrespective of safe_setup decorator used or not in
specific test class. Which results to lot of instance of AttributeError
raised from tearDownclass. For example when tests case is skipped.

This patch provides a workaround by logging AttributeError as 'info'
instead of 'exception'.

Permanent fix for cleaning up those in more appropriate way will
be by split-up resource_setup & resource_cleanup in more structural way
as part of second phase of resource cleanup BP.

Change-Id: Ia49c9e171338e22f8d605fc53d251ef3d3471bb1

tempest/test.py

diff --git a/tempest/test.py b/tempest/test.py
index 4a22b1b..178ca10 100644
--- a/tempest/test.py
+++ b/tempest/test.py
@@ -299,7 +299,14 @@
             try:
                 cls.tearDownClass()
             except Exception as te:
-                LOG.exception("tearDownClass failed: %s" % te)
+                tetype, _, _ = sys.exc_info()
+                # TODO(gmann): Till we split-up resource_setup &
+                # resource_cleanup in more structural way, log
+                # AttributeError as info instead of exception.
+                if tetype is AttributeError:
+                    LOG.info("tearDownClass failed: %s" % te)
+                else:
+                    LOG.exception("tearDownClass failed: %s" % te)
             try:
                 raise etype(value), None, trace
             finally: