https://logs.chromium.org/v/?s=chromium%2Fbb%2Fclient.v8.fyi%2FAndroid_Release__Nexus_5X_%2F2794%2F%2B%2Frecipes%2Fsteps%2Fcompile_confirm_no-op%2F0%2Fstdout lists a bunch of ninja explains about paths being dirty, starting:

ninja explain: output ../../build/android/ant/empty/res/.keep of phony edge with no inputs doesn't exist
ninja explain: ../../build/android/ant/empty/res/.keep is dirty

Andrew, could this be related to r509785?

I'm sure that's it. It was reverted for another reason here:
https://chromium-review.googlesource.com/727902

I'll make sure this doesn't break again in the reland.

If there's anything that I can help with related to the configuration of these waterfall bots let me know.

Looking at the configuration of these bots in:
https://cs.chromium.org/chromium/build/scripts/slave/recipe_modules/chromium_tests/client_v8_fyi.py

they pull top-of-tree V8 but not top-of-tree ANGLE as on the chromium.gpu.fyi waterfall.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/3a77fd434a9935c7e1c4e34e8aaa78aa50582d50

commit 3a77fd434a9935c7e1c4e34e8aaa78aa50582d50
Author: Andrew Grieve <agrieve@chromium.org>
Date: Thu Oct 19 20:31:40 2017

Reland: Android: Remove last remnants of ANT (part 1)

This reverts commit 0462fc40d989e9ed29383e9f183311ea983dac20.

Reason for reland: Leaving keystore and empty/ until recipes
updated and bots build.

TBR=thakis

Bug:  375324 ,776328, 776351 
Change-Id: I8c8cfc0b94ce681772c3792799011e77d72d81d9
Reviewed-on: https://chromium-review.googlesource.com/728460
Commit-Queue: agrieve <agrieve@chromium.org>
Reviewed-by: agrieve <agrieve@chromium.org>
Cr-Commit-Position: refs/heads/master@{#510189}
[delete] https://crrev.com/aeb2dd08d7deec2998db3f0c37843cb1103e9954/build/android/ant/apk-package.xml
[add] https://crrev.com/3a77fd434a9935c7e1c4e34e8aaa78aa50582d50/build/android/chromium-debug.keystore
[add] https://crrev.com/3a77fd434a9935c7e1c4e34e8aaa78aa50582d50/build/android/empty/.keep
[delete] https://crrev.com/aeb2dd08d7deec2998db3f0c37843cb1103e9954/build/android/empty/src/.keep
[modify] https://crrev.com/3a77fd434a9935c7e1c4e34e8aaa78aa50582d50/build/android/gyp/process_resources.py
[modify] https://crrev.com/3a77fd434a9935c7e1c4e34e8aaa78aa50582d50/build/config/android/config.gni
[modify] https://crrev.com/3a77fd434a9935c7e1c4e34e8aaa78aa50582d50/build/config/android/internal_rules.gni
[modify] https://crrev.com/3a77fd434a9935c7e1c4e34e8aaa78aa50582d50/build/config/android/rules.gni
[modify] https://crrev.com/3a77fd434a9935c7e1c4e34e8aaa78aa50582d50/build/install-build-deps-android.sh

What else needs to happen to propagate the fix, since the waterfall is still red? Slave restart? I'll issue it if so, just let me know.

I would have imagined that the revert would be all that's needed.

Doing a clobber build would be the next thing to try.

Triggered one here: https://build.chromium.org/p/client.v8.fyi/builders/Android%20Release%20%28Nexus%205X%29/builds/2839

Still broken :( I'll try restarting the slave.

Looks like the error is now:

ninja explain: restat of output md5sum_bin_host older than most recent input obj/tools/android/md5sum/md5sum_bin_host.inputdeps.stamp (1508459005 vs 1508459008)
ninja explain: md5sum_bin_host is dirty
ninja explain: obj/tools/android/md5sum/md5sum_bin_host.stamp is dirty

This is actually just a known bug with ninja that's stopped us from enabling this same check for the main android bots:
https://bugs.chromium.org/p/chromium/issues/detail?id=646165

It's not been fixed yet though, since I've been unable to construct a repro for the error :(

+thakis fyi. 

FYI: Our bot turned green from https://build.chromium.org/p/client.v8.fyi/builders/Android%20Release%20%28Nexus%205X%29/builds/2842 on.

Note that we use Chromium from heads/lkcr on the Chromium/V8 bots. Hence it takes usually 2 hours longer for commits to propagate.

If it's green, then let's mark this as fixed. I'm guessing it's green more out of coincidence than from anything actually being fixed though. The md5_bin_host being stale has been a long-time flakey issue.

Removing restrict-view-google (I see no reason for it here).

This happens again, same bot. Starting here:
https://build.chromium.org/p/client.v8.fyi/builders/Android%20Release%20%28Nexus%205X%29/builds/3299

Trying a clobber build.

Clobber didn't help:
https://build.chromium.org/p/client.v8.fyi/builders/Android%20Release%20%28Nexus%205X%29/builds/3303

I assume this new instance is duplicate of issue 786876

Sounds likely. https://ci.chromium.org/buildbot/chromium.gpu.fyi/Android%20Release%20%28Nexus%205X%29/?limit=100 is building fine.

Removing 'ninja_confirm_noop' from the definition of "Android Release (Nexus 5X)" tools/build/scripts/slave/recipe_modules/chromium_tests/client_v8_fyi.py will probably work around the problem.

@tikuta: Should we remove the step from the V8 Android GPU fyi bot, too?

I think so if other android builders do not have the step.

But the step looks include some investigatable info.
https://logs.chromium.org/v/?s=chromium%2Fbb%2Fclient.v8.fyi%2FAndroid_Release__Nexus_5X_%2F3299%2F%2B%2Frecipes%2Fsteps%2Fcompile_confirm_no-op%2F0%2Fstdout
ninja explain: expected depfile 'gen/chrome/android/chrome_public_apk__process_resources.d' to mention 'gen/chrome/android/chrome_public_apk__process_resources_R.txt', got 'gen/chrome/android/chrome_public_apk/chrome_public_apk.resources.all.zip'

agrieve@, can you take a look around these files?

Yep, that error is certainly fixable. 

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/3c1a7c69b41b2ff605c61f31c1e9bac3dd2be826

commit 3c1a7c69b41b2ff605c61f31c1e9bac3dd2be826
Author: Andrew Grieve <agrieve@chromium.org>
Date: Tue Nov 21 20:52:36 2017

Android: Fix wrong output listed in process_resources.py depfile

Bug:  776351 
Change-Id: Iecb2d207d5c6383898f1d58e2f0fb2b61f09dbc0
Reviewed-on: https://chromium-review.googlesource.com/782463
Reviewed-by: Eric Stevenson <estevenson@chromium.org>
Commit-Queue: agrieve <agrieve@chromium.org>
Cr-Commit-Position: refs/heads/master@{#518384}
[modify] https://crrev.com/3c1a7c69b41b2ff605c61f31c1e9bac3dd2be826/build/android/gyp/process_resources.py
[modify] https://crrev.com/3c1a7c69b41b2ff605c61f31c1e9bac3dd2be826/build/config/android/internal_rules.gni

The commit should fix the log from the bot, let's leave open until it goes green though.

V8's fyi bot is green again. Thanks for fixing!