Add one more histogram to log how many Request are still pending when

Add one more histogram to log how many Request are still pending when
the job controller count hits the limit: 100, 200, etc.

We have observed potential memory leaks in crbug.com/711721, which
indicates the leak of main job when racing logic is disabled. If the
Request is finished and destructed, the main job must be cleaned up in
OnRequestComplete(). We used to assume the Request will abort if
timed out, maybe there are corner cases where Request will leak.

BUG=704988
Review-Url: https://codereview.chromium.org/2824513003
Cr-Commit-Position: refs/heads/master@{#464847}
Committed: https://chromium.googlesource.com/chromium/src/+/c77104dfbdbdb4e4952dad12fa2d2352a07e65ff

[Zhongyi Shi, 2017-04-14 22:20:27]

zhongyi@chromium.org changed reviewers:
+ rch@chromium.org, xunjieli@chromium.org

[Ryan Hamilton, 2017-04-14 22:22:28]

The CQ bit was checked by rch@chromium.org

[Ryan Hamilton, 2017-04-14 22:22:28]

lgtm

[commit-bot: I haz the power, 2017-04-14 22:22:43]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Zhongyi Shi, 2017-04-14 22:24:25]

zhongyi@chromium.org changed reviewers:
+ holte@chromium.org

[Zhongyi Shi, 2017-04-14 22:24:26]

+ holte@
Steven, could you take a look? If we can get some data this weekend, it will
help us investigate crbug.com/711721.

Thanks a lot!

[commit-bot: I haz the power, 2017-04-14 22:31:07]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-04-14 22:31:08]

Try jobs failed on following builders:
  chromium_presubmit on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presub...)

[Steven Holte, 2017-04-14 22:46:44]

lgtm

https://codereview.chromium.org/2824513003/diff/1/net/http/http_stream_factor...
File net/http/http_stream_factory_impl.cc (right):

https://codereview.chromium.org/2824513003/diff/1/net/http/http_stream_factor...
net/http/http_stream_factory_impl.cc:399:
UMA_HISTOGRAM_COUNTS_1M("Net.JobControllerSet.CountOfPendingRequest",
It seems like 1M is much bigger than range of actual values you are recording in
this histogram, so you might consider using a smaller range.  (This will make
the bucket intervals smaller).

[Zhongyi Shi, 2017-04-14 22:53:24]

On 2017/04/14 22:46:44, Steven Holte wrote:
> lgtm
Thanks for the quick review!

>
https://codereview.chromium.org/2824513003/diff/1/net/http/http_stream_factor...
> File net/http/http_stream_factory_impl.cc (right):
> 
>
https://codereview.chromium.org/2824513003/diff/1/net/http/http_stream_factor...
> net/http/http_stream_factory_impl.cc:399:
> UMA_HISTOGRAM_COUNTS_1M("Net.JobControllerSet.CountOfPendingRequest",
> It seems like 1M is much bigger than range of actual values you are recording
in
> this histogram, so you might consider using a smaller range.  (This will make
> the bucket intervals smaller).

We could reduce the bucket range. Given that we haven't sort out what's the
maximum number we could have if we have aggressive leak (for the time being, the
largest number we saw is ~5k, that number might climbed up as user run Chrome
longer). So I want to keep as is, when the histogram stabilizes and leaks are
cleaned, we could change to a smaller range.

[Zhongyi Shi, 2017-04-14 22:53:30]

The CQ bit was checked by zhongyi@chromium.org

[commit-bot: I haz the power, 2017-04-14 22:53:52]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-04-15 01:43:22]

CQ is committing da patch.

Bot data: {"patchset_id": 1, "attempt_start_ts": 1492210410045230, "parent_rev":
"bc1c102b111b5107384ff8a2cbf806870966a49c", "commit_rev":
"c77104dfbdbdb4e4952dad12fa2d2352a07e65ff"}

[commit-bot: I haz the power, 2017-04-15 01:44:23]

Description was changed from

==========
Add one more histogram to log how many Request are still pending when
the job controller count hits the limit: 100, 200, etc.

We have observed potential memory leaks in crbug.com/711721, which
indicates the leak of main job when racing logic is disabled. If the
Request is finished and destructed, the main job must be cleaned up in
OnRequestComplete(). We used to assume the Request will abort if
timed out, maybe there are corner cases where Request will leak.

BUG=704988
==========

to

==========
Add one more histogram to log how many Request are still pending when
the job controller count hits the limit: 100, 200, etc.

We have observed potential memory leaks in crbug.com/711721, which
indicates the leak of main job when racing logic is disabled. If the
Request is finished and destructed, the main job must be cleaned up in
OnRequestComplete(). We used to assume the Request will abort if
timed out, maybe there are corner cases where Request will leak.

BUG=704988

Review-Url: https://codereview.chromium.org/2824513003
Cr-Commit-Position: refs/heads/master@{#464847}
Committed:
https://chromium.googlesource.com/chromium/src/+/c77104dfbdbdb4e4952dad12fa2d...
==========

[commit-bot: I haz the power, 2017-04-15 01:44:24]

Committed patchset #1 (id:1) as
https://chromium.googlesource.com/chromium/src/+/c77104dfbdbdb4e4952dad12fa2d...