Try swarming base_unittests, browser_tests, content_browsertests, interactive_ui_tests, sync_integr…

Try swarming base_unittests, browser_tests, content_browsertests, interactive_ui_tests, sync_integration_tests and unit_tests on debug Linux bots.

BUG=414808
TBR=maruel@chromium.org

Committed: https://crrev.com/a7a6eca707e8ffa38635eb425fd22169c75c4fd9
Cr-Commit-Position: refs/heads/master@{#296409}

[jam, 2014-09-24 14:37:40]

Patchset #3 (id:40001) has been deleted

[jam, 2014-09-24 14:37:44]

Patchset #2 (id:20001) has been deleted

[jam, 2014-09-24 14:38:42]

If this sticks, I'll remove the builder/tester split and then do the same for
release.

[jam, 2014-09-24 14:41:56]

Committed patchset #2 (id:60001) manually as a7a6eca (presubmit successful).

[commit-bot: I haz the power, 2014-09-24 14:41:59]

Patchset 2 (id:??) landed as
https://crrev.com/a7a6eca707e8ffa38635eb425fd22169c75c4fd9
Cr-Commit-Position: refs/heads/master@{#296409}

[M-A Ruel, 2014-09-24 14:45:57]

On 2014/09/24 14:38:42, jam wrote:
> If this sticks, I'll remove the builder/tester split and then do the same for
> release.

We'll wait for a few stability features to be implemented in Swarming before
builder/tester split removal. Keeping this setup makes it possible to revert
back in case of catastrophic failure.

[M-A Ruel, 2014-09-24 14:46:05]

lgtm

[jam, 2014-09-24 14:51:21]

On 2014/09/24 14:45:57, M-A Ruel wrote:
> On 2014/09/24 14:38:42, jam wrote:
> > If this sticks, I'll remove the builder/tester split and then do the same
for
> > release.
> 
> We'll wait for a few stability features to be implemented in Swarming before
> builder/tester split removal.

oh, which stability features were you thinking of?
From my pov, the ones I was thinking of were a) reboot occasionally and b) retry
tasks with BOT_DIED, both of which were done.

> Keeping this setup makes it possible to revert
> back in case of catastrophic failure.

That's true; I had thought that worst case scenario we can just disable swarming
with one machine running tests. It will be slow, but given that we've been using
swarming for 2 months now, it seems very low risk. Also, if there is a big
failure, CQ being down seems just as bad as main waterfall, and either way cls
won't be landing. I don't feel strongly if you want to keep the machines for
now, but just wanted to let you know what was my reasoning.