Use rate of 5 to avoid false failures

See the comment in the file for more details

Change-Id: I4a54fbc693747aa0a6d49e56700a9e61b25fe99d
Signed-off-by: Jamo Luhrsen <jluhrsen@redhat.com>

diff --git a/csit/suites/controller/cluster_singleton/chasing_the_leader.robot b/csit/suites/controller/cluster_singleton/chasing_the_leader.robot
index ad7620b4beb00defe7ac9bc6c4b115aff6c5cf78..63e83526aa7425a2a7934185b566a9774315f0ad 100644 (file)
--- a/csit/suites/controller/cluster_singleton/chasing_the_leader.robot
+++ b/csit/suites/controller/cluster_singleton/chasing_the_leader.robot
@@ -21,8 +21,13 @@ Resource          ${CURDIR}/../../../libraries/SetupUtils.robot
 
 *** Variables ***
 ${TEST_DURATION}    1m
-${ACCEPTED_PER_SEC_RATE}    10
-# TODO: Use a better wait to start testing when ODL does not use CPU for initializing features. Then inrease rate to 100.
+${ACCEPTED_PER_SEC_RATE}    5
+# NOTE: The shared infra that we use can sometimes be sluggish and we get false failures.
+# There are cases when the rate will be under 20 as well as over 100. Normally it is over
+# 50, but in order to stabilize our tests and avoid false failures this rate has been set
+# low. The test will still fail if there are failed registrations during the "flapping".
+# Some discussion of this can be found in the comments of
+# https://git.opendaylight.org/gerrit/#/c/74692/.
 
 *** Test Cases ***
 Register_Candidates