[Swan-commit] Changes to ref refs/heads/master

[Paul Wouters]

New commits:
commit 43e1428e8c5b070b2dd109a99ad3a4c718a8cacc
Author: Paul Wouters <pwouters at redhat.com>
Date:   Wed Feb 27 23:45:00 2013 -0500

    * testing: Figured out the occasional mangling of lines!!
    
    The cause was that we were waiting in the expect loop on the prompt
    to return, to then send the next line. But the prompt was defined
    as "root at hostname", even though it was "[root at hostname testname]# ".
    
    probably the [] were left out because within expect that also has
    meaning so you have to protect them using \[ and \]
    
    With matching the full prompt now, I managed to run basic-pluto-01
    5 times in a row without seeing the mangling anywhere.

commit eee8e35e170f32d9d9a568f141bb76668c660c8a
Author: Paul Wouters <pwouters at redhat.com>
Date:   Wed Feb 27 23:44:14 2013 -0500

    * testing: north gets a new raw rsa key that's in NSS

commit aad9f13140ed57b2c6f3fccb85682d0226d390fb
Author: Paul Wouters <pwouters at redhat.com>
Date:   Wed Feb 27 23:14:45 2013 -0500

    * testing: remove unused virtinstall-base

commit 3dcf525c51d81c44b88bd389bc74fc2e671d05b5
Author: Paul Wouters <pwouters at redhat.com>
Date:   Wed Feb 27 22:59:58 2013 -0500

    * testing: dotest.sh now logs results to testname/RESULT

commit ba895127bf6fa79d5f37d8b522f0d577b81aa24e
Author: Paul Wouters <pwouters at redhat.com>
Date:   Wed Feb 27 21:52:30 2013 -0500

    * testing: Make dotest.sh and runkvm.py a little more robust
    
    I was regularly seeing "hangs" and a failure for a test case to take
    control of a VM. These changes resolve the hangs for me and improves
    the output for humans to figure out what's going on.
    
    We used the serial console to reboot a VM, now we use virsh reboot
    directly. When we still had something running (eg ipsec auto --up retrying
    a long time) our reboot command would never arrive.
    
    Related, we now hit return and ctrl-c when reconnecting to the
    VM. Additionally, we run stty sane because the console's tty insanity upon
    reconnect seems to cause what we called "flow problems" in the output.
    
    Use setproctitle to rename the runkvm.py processes (called python) to
    "swankvm". This allows us to "killall swankvm" at the start of a test
    to kill any lingering python scripts from previous runs. We also kill
    any remaining tcpdump processes.
    
    Prepend the prompt (hostname at testname:) for all output to the shell
    running the test, to make it easier for the human to see which of the
    running hosts is generating the output while the test is running. This
    does not change the output in the test OUTPUT/ directory
    
    wrap all child.expect() calls into a try: / except: statement, so we can
    just throw a human readable error, instead of a python stack trace that
    scrolls off the screen, especially when sharing a screen with 'screen'.
    
    When expect is waiting on either the login: prompt or the root prompt,
    act differently based on which we actually get back. Only attempt to
    login when we did not get a root prompt.
    
    Move deletion of /tmp/pluto.log and symlink from runkvm.py to swan-prep
    
    Also reboot "nic" for each test so it properly clears the iptables and
    conntrack tables.
    
    Reduce the timeout values for expect so failing tests fail a little quicker.
    Before this it would take minutes to fail.
    
    Clearly notify failure/success for gaining access to a VM.
    
    When hitting return to get a shell prompt, also attempt ctrl-c