BSC_Tests: Avoid race condition between paging cmd and reset ack
Sometimes (under heavy load and when the last paging cmd arrives near
the reset ack) some messages are not enqueued in the IPA_RSL port after
we have received the reset ack from BSSAP.
In the failures I have seen wireshark reports that no paging cmd arrived
after the reset ack, see jenkins job 285:
https://jenkins.osmocom.org/jenkins/view/TTCN3/job/ttcn3-bsc-test/285/
Adding a sleep is always a bit crude, but since the paging cmd is
scheduled through TCP while the reset ack is sent through sctp I don't
see another way.
Running the test under load in a loop showed improvements where I was
not able to reproduce the failure with this patch.
Change-Id: Ib2d60e2c59baf98e437e078d844adcc6dbdbfcd8
diff --git a/bsc/BSC_Tests.ttcn b/bsc/BSC_Tests.ttcn
index c802a6b..5af4573 100644
--- a/bsc/BSC_Tests.ttcn
+++ b/bsc/BSC_Tests.ttcn
@@ -1390,6 +1390,10 @@
[] BSSAP.receive { repeat; }
}
+ /* Wait to avoid a possible race condition if a paging message is
+ * received right before the reset ACK. */
+ f_sleep(0.2);
+
/* Clear the queue, it might still contain stuff like BCCH FILLING */
for (var integer i := 0; i < sizeof(IPA_RSL); i := i+1) {
IPA_RSL[i].clear;