Ensure disk cache init to make PCV1_COLD results stable

Ensure disk cache init to make PCV1_COLD results stable

Before this CL, the first PCV1_COLD run after browser launch was slightly
slower than the subsequent runs, as browser process DiskCache init was
still running.
This CL workarounds the issue by forcing a navigation to "example.com"
website before the first PCV1_COLD page test run.

BUG=catapult:#2342

Committed: https://chromium.googlesource.com/external/github.com/catapult-project/catapult/+/afe031977961e1a91dfb8d8a8d2222cc2a222a1d

[kouhei (in TOK), 2016-05-17 05:52:37]

kouhei@chromium.org changed reviewers:
+ eakuefner@chromium.org, nednguyen@google.com

[kouhei (in TOK), 2016-05-17 05:52:38]

PTAL

[nednguyen, 2016-05-17 15:33:36]

On 2016/05/17 05:52:38, kouhei wrote:
> PTAL

Why example.com? That would mean all the pagesets used for page cycler will need
to have example.com in its recording. How does the current page_cycler deal with
this problem?

[kouhei (in TOK), 2016-05-17 22:05:47]

On 2016/05/17 15:33:36, nednguyen wrote:
> On 2016/05/17 05:52:38, kouhei wrote:
> > PTAL
> 
> Why example.com? That would mean all the pagesets used for page cycler will
need
> to have http://example.com in its recording. How does the current page_cycler
deal with
> this problem?

http://example.com doesn't have to exist in its recording and just rendering 404
page is enough, but it has to be a URL which triggers disk cache init.
AFAIK, pcv1 doesn't deal with this problem and just report noisy results.

[nednguyen, 2016-05-17 22:22:23]

On 2016/05/17 22:05:47, kouhei wrote:
> On 2016/05/17 15:33:36, nednguyen wrote:
> > On 2016/05/17 05:52:38, kouhei wrote:
> > > PTAL
> > 
> > Why example.com? That would mean all the pagesets used for page cycler will
> need
> > to have http://example.com in its recording. How does the current
page_cycler
> deal with
> > this problem?
> 
> http://example.com doesn't have to exist in its recording and just rendering
404
> page is enough, but it has to be a URL which triggers disk cache init.
> AFAIK, pcv1 doesn't deal with this problem and just report noisy results.

Can we just use about:blank or http://does.not.exist? :-)

[kouhei (in TOK), 2016-05-17 22:34:42]

On 2016/05/17 22:22:23, nednguyen wrote:
> On 2016/05/17 22:05:47, kouhei wrote:
> > On 2016/05/17 15:33:36, nednguyen wrote:
> > > On 2016/05/17 05:52:38, kouhei wrote:
> > > > PTAL
> > > 
> > > Why example.com? That would mean all the pagesets used for page cycler
will
> > need
> > > to have http://example.com in its recording. How does the current
> page_cycler
> > deal with
> > > this problem?
> > 
> > http://example.com doesn't have to exist in its recording and just rendering
> 404
> > page is enough, but it has to be a URL which triggers disk cache init.
> > AFAIK, pcv1 doesn't deal with this problem and just report noisy results.
> 
> Can we just use about:blank or http://does.not.exist :-)

about:blank doesn't trigger disk cache, but http://does.not.exist sgtm. ack.

[kouhei (in TOK), 2016-05-17 23:45:40]

Description was changed from

==========
Ensure disk cache init to make PCV1_COLD results stable

Before this CL, the first PCV1_COLD run after browser launch was slightly
slower than the subsequent runs, as browser process DiskCache init was
still running.
This CL workarounds the issue by forcing a navigation to "example.com"
website before the first PCV1_COLD page test run.

BUG=None
==========

to

==========
Ensure disk cache init to make PCV1_COLD results stable

Before this CL, the first PCV1_COLD run after browser launch was slightly
slower than the subsequent runs, as browser process DiskCache init was
still running.
This CL workarounds the issue by forcing a navigation to "example.com"
website before the first PCV1_COLD page test run.

BUG=catapult:#2342
==========

[kouhei (in TOK), 2016-05-17 23:46:00]

On 2016/05/17 22:34:42, kouhei wrote:
> On 2016/05/17 22:22:23, nednguyen wrote:
> > On 2016/05/17 22:05:47, kouhei wrote:
> > > On 2016/05/17 15:33:36, nednguyen wrote:
> > > > On 2016/05/17 05:52:38, kouhei wrote:
> > > > > PTAL
> > > > 
> > > > Why example.com? That would mean all the pagesets used for page cycler
> will
> > > need
> > > > to have http://example.com in its recording. How does the current
> > page_cycler
> > > deal with
> > > > this problem?
> > > 
> > > http://example.com doesn't have to exist in its recording and just
rendering
> > 404
> > > page is enough, but it has to be a URL which triggers disk cache init.
> > > AFAIK, pcv1 doesn't deal with this problem and just report noisy results.
> > 
> > Can we just use about:blank or http://does.not.exist :-)
> 
> about:blank doesn't trigger disk cache, but http://does.not.exist sgtm. ack.

Done.

[nednguyen, 2016-05-18 00:04:37]

lgtm

[kouhei (in TOK), 2016-05-18 00:05:49]

The CQ bit was checked by kouhei@chromium.org

[commit-bot: I haz the power, 2016-05-18 00:05:54]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1984973002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1984973002/20001

[commit-bot: I haz the power, 2016-05-18 00:26:51]

Description was changed from

==========
Ensure disk cache init to make PCV1_COLD results stable

Before this CL, the first PCV1_COLD run after browser launch was slightly
slower than the subsequent runs, as browser process DiskCache init was
still running.
This CL workarounds the issue by forcing a navigation to "example.com"
website before the first PCV1_COLD page test run.

BUG=catapult:#2342
==========

to

==========
Ensure disk cache init to make PCV1_COLD results stable

Before this CL, the first PCV1_COLD run after browser launch was slightly
slower than the subsequent runs, as browser process DiskCache init was
still running.
This CL workarounds the issue by forcing a navigation to "example.com"
website before the first PCV1_COLD page test run.

BUG=catapult:#2342

Committed:
https://chromium.googlesource.com/external/github.com/catapult-project/catapu...
==========

[commit-bot: I haz the power, 2016-05-18 00:26:52]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/external/github.com/catapult-project/catapu...

[nednguyen, 2016-05-18 01:51:13]

https://codereview.chromium.org/1984973002/diff/20001/telemetry/telemetry/pag...
File telemetry/telemetry/page/cache_temperature.py (right):

https://codereview.chromium.org/1984973002/diff/20001/telemetry/telemetry/pag...
telemetry/telemetry/page/cache_temperature.py:64:
tab.Navigate("http://does.not.exist")
Err I forgot to this. Can you add some comment explaining why we do this? Future
maintainers will thank us :-)