I think this failure is due to a recent change to this test and the fact that Chromium isn't handling multisampled floating-point renderbuffers correctly yet. I think there's another bug already filed about this. The suppression in webgl2_conformance_expectations.py should be expanded.

Still failing on this bot.
https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Release%20(ATI)

Need to relax the code in Chromium to return a sample count of at least one for these renderbuffer formats, even if that sample count is just 1.

Failing on https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Release%20%28New%20Intel%29?numbuilds=200 with

Traceback (most recent call last):
  _RunGpuTest at content/test/gpu/gpu_tests/gpu_integration_test.py:71
    self.RunActualGpuTest(url, *args)
  RunActualGpuTest at content/test/gpu/gpu_tests/webgl_conformance_integration_test.py:50
    self.fail(webgl_conformance._WebGLTestMessages(self.tab))
  fail at /usr/lib/python2.7/unittest/case.py:410
    raise self.failureException(msg)
AssertionError: floating-point RGBA32F render target not supported
FAIL floating-point RGBA32F render target not supported

That last comment is actually about linux intel; i'll file a new bug. 

No longer seeing this flake on the bot. Is this fixed?

The test's marked failing in webgl2_conformance_expectations.py per  Issue 633022 . I'm going to duplicate this bug into that one.