[CentOS-devel] Fwd: sig-core-t_functional suite for 7.2.1511

Fri Jan 29 14:22:18 UTC 2016
Thierry Fauck@linux.vnet.ibm.com <thierry at linux.vnet.ibm.com>

As talked earlier today,

Here are my comments - I understand they are mostly related to the
environment it is running into, but it may be interesting to look into
it to improve the cases.

Regards
Thierry (nick - thf )


-------- Forwarded Message --------
Subject: sig-core-t_functional suite for 7.2.1511
Date: Tue, 22 Dec 2015 11:14:29 +0100
From: Thierry Fauck at linux.vnet.ibm.com <thierry at linux.vnet.ibm.com>
To: centos-qa at centos.org

Hello,

While checking the suite for ppc64 and ppc64le, I got the following
changes and comment - let me know if you want a diff file.

- runtest.sh
If script fails and you relaunch it a second time, you may get a failure
because tests users are already created - I suggested to remove all test
users in runtests.sh to have a clean environment.

In t_CheckExitStatus shouldn't we add the test name being run to ease
tracing and script debugging like ?
	[ $1 -eq 0 ] && { t_Log "$0 PASS"; return $PASS; }

- Depending on your setting, you may come with LC_ALL setting as utf8 or
C and 10-cmp-tests may fails - checking for byte or char would allow
both settings to be run.
	cmp -i 15:16 ${FILE}a ${FILE}b | egrep -q "byte 1, line 1|char 1, line
1"|| ExitFail

- tests/p_mtr/mtr_test.sh fails if you are not in the centos.org env as
it tries to reach the qa host. As this test is done at start of
runtest.sh, in case it is not true, shouldn't you use another hosts
found on your local network (an easy bypass is to use like the local DNS
server or the gateway ?).

- p_shadow-utils/20-chage_tests may fails because of capital letters jan
versus Jan - I would suggest to use
	chage -l testshadow | grep Last | grep -qi "Jan 01, 2013"

- tests/p_shim/01_shim_secureboot_signed.sh
  tests/p_kernel/02_kernel_secureboot_signed.sh
	are x86_64 dependant

- Had to increase -T delay from 2 to 8 in tests/p_squid/squid_test.sh

- tests/p_systemd/25-journalctl.sh - for fast processors, you need to
had a delay for journalctl to cope with your message - I added a sleep 1
after writting to /dev/kmsg

- tests/p_which/which_basic.sh - sounds like wich is now in /bin instead
of /usr/bin - /usr/bin/which which | grep -q "/bin/which" would cope
with both cases

Hope that helps

-- 
Thierry