Chrome Version: (copy from chrome://version)
OS: (e.g. Win7, OSX 10.9.5, etc...)

What steps will reproduce the problem?
(1) unit_tests with --test-launcher-retry-limit=0

What is the expected result?

Expect that all the tests pass, every time.

What happens instead?

On my workstation[1], ~3 tests fail and ~70-75 actually crash; I'll file separate bugs for burning down those flakes.

It's easy for these regressions to creep in because we have a default test-launcher retry-limit of 3 times.

For the vast majority of unit-tests, there is no legitimate reason for flake - we should be able to switch those tests to run without retries - but some tests are inherently fragile e.g. due to pesky operating system edge-case conditions.

To allow for tests which are valuable most of the time but inevitably flake once in a while, we could add an annotation to the test name (remember our old friend FLAKY_, anyone), which would tell the test-launcher to allow that individual test to be retried, and disallow retries for all other tests.

e.g. if we named the annotation EVIL_RETRY_HACK_ then we might have:

TEST(ExtensionServiceTest, LovelySimpleTest)
TEST(ExtensionServiceTest, EVIL_RETRY_HACK_TestThatReliesOnSystemResourceAndSometimesFailsOnXP)