Restore the persistent server option in the LB tests
The cb85dcdf5e2cf53ac21bb97382e03a8c5b47042f putted the
nc command into loop in order to tolerate busybox/nc versions
without persistent server option (cirros 0.3.3).
Unfortunately the above change also changed -ll option to -l.
The nc with -ll option provides 'never stop listening' service,
without this behavior the test can fail frequently/always
in several setup.
This change restores the -ll option, which does not causes issues
when you use cirros 0.3.3 image, and gives stability with cirros 0.3.2
+ haproxy 1.5.x.
Note: The gate is using 0.3.2 cirros and HaProxy 1.4.x,
On the gate this change effectively restores the behavior before
cb85dcdf5e2cf53ac21bb97382e03a8c5b47042f.
Change-Id: I87bee99946447dee499ab3b8ca2b1cacb54668fd
Related-Bug: #1408922
Related-Bug: #1384708
diff --git a/tempest/scenario/test_load_balancer_basic.py b/tempest/scenario/test_load_balancer_basic.py
index 57a5406..2516a47 100644
--- a/tempest/scenario/test_load_balancer_basic.py
+++ b/tempest/scenario/test_load_balancer_basic.py
@@ -185,7 +185,7 @@
# Start netcat
start_server = ('while true; do '
- 'sudo nc -l -p %(port)s -e sh /tmp/%(script)s; '
+ 'sudo nc -ll -p %(port)s -e sh /tmp/%(script)s; '
'done &')
cmd = start_server % {'port': self.port1,
'script': 'script1'}