So, I think this is a bad GN or similar change landing into chromium around some time before midnight MTV time (~9AM for Munich). I base this on when first failing builds started appearing on 3 builders referenced above.

https://build.chromium.org/p/chromium.gpu.fyi/builders/GPU%20Win%20Builder%20%28dbg%29?numbuilds=200

Aug 08 00:37	242e2067b136691816302d52dd63ac1614f9d843	exception	#59452
Exception steps failed isolate tests
Aug 08 00:20	8f51b9de091614f73bf63e53be547a12d980508c	failure	#59451
Failed steps failed compile confirm no-op failed failure reason
Aug 07 23:52	034f09fb84756039a913e27ceba7556ad8ffb891	success	#59450
Build successful


https://build.chromium.org/p/chromium.gpu/builders/GPU%20Win%20Builder%20%28dbg%29?numbuilds=200

Aug 08 00:33	242e2067b136691816302d52dd63ac1614f9d843	exception	#59900
Exception steps failed isolate tests
Aug 08 00:01	e06d40f52c4a69790bfb31deaea6595b8443fd2c	failure	#59899
Failed steps failed compile confirm no-op failed failure reason
Aug 07 23:35	19a4ec7b418d9672991c68057bbb084edbbfd76f	success	#59898
Build successful



https://build.chromium.org/p/chromium.gpu.fyi/builders/GPU%20Win%20x64%20Builder%20%28dbg%29?numbuilds=200

Aug 08 00:23	8f51b9de091614f73bf63e53be547a12d980508c	exception	#22280
Exception steps failed isolate tests
Aug 07 23:47	034f09fb84756039a913e27ceba7556ad8ffb891	success	#22279
Build successful

Looking at commit positions, 


last green builds blamelists:

492556
492553
492556

first purple builds:
492567
492567
492563


=> culprit should be in this inclusive range 492557..492563

https://chromium.googlesource.com/chromium/src/+log/8f51b9de091614f73bf63e53be547a12d980508c~7..8f51b9de091614f73bf63e53be547a12d980508c

Most likely are https://chromium.googlesource.com/chromium/src/+/ebe22580ea36ba92cf08d27d4137fa7a9bf39e3a

pdf/pdfium/fuzzers/BUILD.gn[diff]
testing/libfuzzer/archive_corpus.py[diff]
testing/libfuzzer/efficient_fuzzer.md[diff]
testing/libfuzzer/fuzzers/BUILD.gn[diff]
third_party/WebKit/Source/platform/BUILD.gn[diff]


OR 

https://chromium.googlesource.com/chromium/src/+/8f51b9de091614f73bf63e53be547a12d980508c


content/child/service_worker/service_worker_dispatcher.cc[diff]
content/child/service_worker/service_worker_dispatcher.h[diff]
content/child/service_worker/service_worker_dispatcher_unittest.cc[diff]
content/child/service_worker/service_worker_network_provider.cc[diff]
content/child/service_worker/service_worker_provider_context.cc[diff]
content/child/service_worker/service_worker_provider_context.h[diff]
content/child/service_worker/service_worker_provider_context_unittest.cc[Added - diff]
content/child/service_worker/service_worker_registration_handle_reference.h[diff]
content/renderer/service_worker/service_worker_context_client.cc[diff]
content/renderer/service_worker/service_worker_context_client.h[diff]
content/test/BUILD.gn[diff]
third_party/WebKit/Source/modules/exported/WebEmbeddedWorkerImpl.cpp[diff]
third_party/WebKit/Source/modules/serviceworkers/WebEmbeddedWorkerImplTest.cpp[diff]
third_party/WebKit/public/web/modules/serviceworker/WebServiceWorkerContextClient.h[diff]

+ Authors of two above CLs
+ dpranke@ who probably understands how isolate step gets these paths and why they don't exist.

I've triggered two builds on two specific revisions  on this builder https://uberchromegw.corp.google.com/i/chromium.gpu.fyi/builders/GPU%20Win%20Builder%20%28dbg%29 to help bisect:

ebe22580ea36ba92cf08d27d4137fa7a9bf39e3a
034f09fb84756039a913e27ceba7556ad8ffb891

Hm, I don't believe that my CL (https://chromium.googlesource.com/chromium/src/+/ebe22580ea36ba92cf08d27d4137fa7a9bf39e3a) could affect that, as all my changes were inside of fuzzer_test targets, which are linux only: https://cs.chromium.org/chromium/src/testing/libfuzzer/fuzzer_test.gni

And the changes were trivial: fixing non existing paths into existing (e.g. "src/pdf/something" -> "//pdf/something", etc.)

This is likely a duplicate of  issue 753774 .

 Issue 753368  has been merged into this issue.

 Issue 753788  has been merged into this issue.

+ infra troopers as FYI. I don't think this is a trooper bug. Instead, it's for sheriffs to figure out what's wrong.

And two rebuilds i've scheduled resulted in the same error, including the 034f09fb84756039a913e27ceba7556ad8ffb891 which used to be green. Hence, my culprit analysis is likely not so useful since incremental building looks broken.

Clobbering builds might help, but it's still unclear what's the culprit CL.

To clarify with people unfamiliar with the system:

There's a BUILD.gn target that states that blink_web.dll.pdb shall be generated and mapped in for a test but it is not built.

Bug is at:
https://cs.chromium.org/chromium/src/tools/perf/chrome_telemetry_build/BUILD.gn?l=65

Thanks to Alexis for finding. A sheriff needs to update this BUILD.gn.

#14: ack, working on it

+nverne, as this is almost certainly related to https://chromium.googlesource.com/chromium/src/+/8ef90c60c92c7ce9283906160e2dbeaf384cca40 or https://chromium.googlesource.com/chromium/src/+/e6c400f69b826b7b30f9d5527ae3862fce47124b. Former looks more likely but the timing is off.

...and, given the following part of the commit description:

"Rehome remaining STATIC_ASSERT_ENUMs. This leaves blink_web empty. Since
Windows doesn't allow empty libraries, we need to remove dependencies on
blink's web/ lib."

it seems like it's safe to simply remove that line form chrome_telemetry_build.

fix in the CQ: https://chromium-review.googlesource.com/c/608611

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/7c9823c1e2dab5bf52c6191531c397efb2947965

commit 7c9823c1e2dab5bf52c6191531c397efb2947965
Author: John Budorick <jbudorick@chromium.org>
Date: Wed Aug 09 16:58:14 2017

Remove blink_web from telemetry_chrome_test data.

TBR=nednguyen@chromium.org

Bug:  753774 
Change-Id: I82c76d3d48cce6cf2c4488b457fd753c7a26fac7
Reviewed-on: https://chromium-review.googlesource.com/608611
Reviewed-by: John Budorick <jbudorick@chromium.org>
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Commit-Queue: John Budorick <jbudorick@chromium.org>
Cr-Commit-Position: refs/heads/master@{#493027}
[modify] https://crrev.com/7c9823c1e2dab5bf52c6191531c397efb2947965/tools/perf/chrome_telemetry_build/BUILD.gn

-Troopers

If theres anything the trooper can help with please add the trooper label back.

Should be fixed after #20.

er, #19

Thanks for taking care of this. I was relying on the default bots used by Gerrit's commit queue and didn't realize there were other dependencies on blink_web still lurking.