Passes cache location to tests and runs the tests in parallel.

Passes cache location to tests and runs the tests in parallel.

As a warning, this is a pretty big change.  At a high-level,
this changes the harness to move the managing of the devserver from
image_to_live into the actual test harness.  Paths of the cache locations (for
archive_url) are taken when pre-generating the updates and stored
in a dictionary (maps "path_to_base->path_to_target" (or path for full updates)->
cache paths).

This change also has the tests run in parallel.  Because we now start
X number of VM's at once, each VM needs it's own pid file and ssh_port.
This logic was added as well as running the actual tests in parallel.

Change-Id: I1275d79740c50c2a8028489b43dcbbcf5bbd56c4

BUG=chromium-os:10723
TEST=Ran it ... a lot with -q but without a test_prefix (so full test suite).

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=c418a8f

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=be787f3

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=f2ee5f6

[sosa, 2011-01-25 01:53:35]

Sorry didn't mean to Close this.  PTAL

[dgarrett, 2011-01-25 03:43:54]

On 2011/01/25 01:53:35, sosa wrote:
> Sorry didn't mean to Close this.  PTAL

LGTM with a few nits.

[dgarrett, 2011-01-25 03:44:01]

http://codereview.chromium.org/6277015/diff/3001/bin/cros_au_test_harness.py
File bin/cros_au_test_harness.py (right):

http://codereview.chromium.org/6277015/diff/3001/bin/cros_au_test_harness.py#...
bin/cros_au_test_harness.py:561: cmd = ['%s/cros_run_vm_update' %
self.crosutilsbin,
Indention is off.

http://codereview.chromium.org/6277015/diff/3001/bin/cros_au_test_harness.py#...
bin/cros_au_test_harness.py:572: cmd = ['%s/cros_run_vm_update' %
self.crosutilsbin,
More indention is off.

http://codereview.chromium.org/6277015/diff/3001/bin/cros_au_test_harness.py#...
bin/cros_au_test_harness.py:674: """Small wrapper for threading.Thread that
releases a semaphores on exit."""
Space after period.

http://codereview.chromium.org/6277015/diff/3001/bin/cros_au_test_harness.py#...
bin/cros_au_test_harness.py:684: ending_semaphore:  Semaphore is released every
time a job ends.  Should be
I think it's only one space after the ':', and sure it's only one after the '.'s

http://codereview.chromium.org/6277015/diff/3001/bin/cros_au_test_harness.py#...
bin/cros_au_test_harness.py:700: """Thread override.  Runs the method specified
and sets output."""
Two spaces after period.

[sosa, 2011-01-26 02:31:58]

PTAL ... I've had to change some stuff around cause of the sudo issue ...
specifically look @ diff 4 in the au test harness (changed to use RunCommand
with enter_chroot + sudo and changed kill).

[sosa, 2011-01-26 02:32:25]

(FYI re-tested with Simple and 12 items already owned by root in my cache)

[sosa, 2011-01-26 19:02:02]

@Petkov can you take a look. Don might be out sick.

[dgarrett, 2011-01-26 19:10:56]

On 2011/01/26 19:02:02, sosa wrote:
> @Petkov can you take a look. Don might be out sick.

I'm back in today, looking now.

[dgarrett, 2011-01-26 19:18:56]

On 2011/01/26 19:10:56, dgarrett wrote:
> On 2011/01/26 19:02:02, sosa wrote:
> > @Petkov can you take a look. Don might be out sick.
> 
> I'm back in today, looking now.

Looks good, but I'm not sure I understand why this is needed.

[dgarrett, 2011-01-26 19:19:16]

http://codereview.chromium.org/6277015/diff/24001/bin/cros_au_test_harness.py
File bin/cros_au_test_harness.py (right):

http://codereview.chromium.org/6277015/diff/24001/bin/cros_au_test_harness.py...
bin/cros_au_test_harness.py:735: RunCommand(['sudo',
Should this also be error_ok, to prevent nasty logging when the pkill below is
used?

[sosa, 2011-01-26 19:22:18]

http://codereview.chromium.org/6277015/diff/24001/bin/cros_au_test_harness.py
File bin/cros_au_test_harness.py (right):

http://codereview.chromium.org/6277015/diff/24001/bin/cros_au_test_harness.py...
bin/cros_au_test_harness.py:735: RunCommand(['sudo',
I thought maybe too ... but it seems that cherrypy handles the kill signal well
and ends with exit 0.

Here's the end output from my test run:

[25/Jan/2011:18:30:56] ENGINE Caught signal SIGTERM.
[25/Jan/2011:18:30:56] ENGINE Bus STOPPING
[25/Jan/2011:18:30:56] ENGINE HTTP Server
cherrypy._cpwsgi_server.CPWSGIServer(('0.0.0.0', 8080)) shut down
[25/Jan/2011:18:30:56] ENGINE Stopped thread 'Autoreloader'.
[25/Jan/2011:18:30:56] ENGINE Stopped thread '_TimeoutMonitor'.
[25/Jan/2011:18:30:56] ENGINE Bus STOPPED
[25/Jan/2011:18:30:56] ENGINE Bus EXITING
[25/Jan/2011:18:30:56] ENGINE Bus EXITED
[25/Jan/2011:18:30:56] ENGINE Waiting for child threads to terminate...
INFO   : Unmounting chroot environment.



On 2011/01/26 19:19:16, dgarrett wrote:
> Should this also be error_ok, to prevent nasty logging when the pkill below is
> used?