Log UMA stats on Typed URL DB errors instead of stopping sync.

Log UMA stats on Typed URL DB errors instead of stopping sync.

BUG=123219
TEST=existing tests suffice


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=134709

[Andrew T Wilson (Slow), 2012-04-28 00:35:15]

Roger, PTAL. Anyone else on the sync team could look at this instead if you have
cycles before the offsite.

[Nicolas Zea, 2012-04-28 00:50:26]

LGTM with some comments.

https://chromiumcodereview.appspot.com/10257026/diff/4001/chrome/browser/sync...
File chrome/browser/sync/glue/typed_url_change_processor.cc (right):

https://chromiumcodereview.appspot.com/10257026/diff/4001/chrome/browser/sync...
chrome/browser/sync/glue/typed_url_change_processor.cc:84:
model_associator_->GetErrorPercentage());
Do you care about recording the cases where the percentage is zero? I imagine
those will (should) be the vast majority. perhaps only record if non zero? (here
and elsewhere)

https://chromiumcodereview.appspot.com/10257026/diff/4001/chrome/browser/sync...
File chrome/browser/sync/profile_sync_service_typed_url_unittest.cc (right):

https://chromiumcodereview.appspot.com/10257026/diff/4001/chrome/browser/sync...
chrome/browser/sync/profile_sync_service_typed_url_unittest.cc:863:
syncable::TYPED_URLS));
check that error percentage/error count is non-zero as well? (here and below)

[Andrew T Wilson (Slow), 2012-04-28 01:04:36]

https://chromiumcodereview.appspot.com/10257026/diff/4001/chrome/browser/sync...
File chrome/browser/sync/glue/typed_url_change_processor.cc (right):

https://chromiumcodereview.appspot.com/10257026/diff/4001/chrome/browser/sync...
chrome/browser/sync/glue/typed_url_change_processor.cc:84:
model_associator_->GetErrorPercentage());
On 2012/04/28 00:50:26, nzea wrote:
> Do you care about recording the cases where the percentage is zero? I imagine
> those will (should) be the vast majority. perhaps only record if non zero?
(here
> and elsewhere)
We want to record for every client. My goal is to get two pieces of information:
how many clients get errors, and for the clients who get errors, what percentage
of their DB accesses result in errors (really, I'm expecting this to be either
"almost 100%" or "<5%"). So we need to record zero percentages so I can
determine how common errors are.

https://chromiumcodereview.appspot.com/10257026/diff/4001/chrome/browser/sync...
File chrome/browser/sync/profile_sync_service_typed_url_unittest.cc (right):

https://chromiumcodereview.appspot.com/10257026/diff/4001/chrome/browser/sync...
chrome/browser/sync/profile_sync_service_typed_url_unittest.cc:863:
syncable::TYPED_URLS));
On 2012/04/28 00:50:26, nzea wrote:
> check that error percentage/error count is non-zero as well? (here and below)

Sadly, I couldn't find an easy way to do this since we don't have a reference to
the model associator. But I'll poke around - maybe I can read this directly out
of the UMA var...

[commit-bot: I haz the power, 2012-05-01 15:07:15]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/atwilson@chromium.org/10257026/19002

[commit-bot: I haz the power, 2012-05-01 16:41:44]

Change committed as 134709