Issue metadata
Sign in to add a comment
|
Trybot mac_optional_gpu_tests_rel is failing tests all the time |
||||||||||||||||||||||||
Issue descriptionhttps://build.chromium.org/p/tryserver.chromium.mac/builders/mac_optional_gpu_tests_rel/builds/8166 is repeatably failing * gpu_tests.webgl_conformance_integration_test.WebGLConformanceIntegrationTest.WebglConformance_conformance2_textures_webgl_canvas_tex_2d_r11f_g11f_b10f_rgb_unsigned_int_10f_11f_11f_rev * gpu_tests.webgl_conformance_integration_test.WebGLConformanceIntegrationTest.WebglConformance_conformance2_textures_webgl_canvas_tex_2d_r16f_red_half_float * gpu_tests.webgl_conformance_integration_test.WebGLConformanceIntegrationTest.WebglConformance_conformance2_textures_webgl_canvas_tex_2d_rgb5_a1_rgba_unsigned_byte * gpu_tests.webgl_conformance_integration_test.WebGLConformanceIntegrationTest.WebglConformance_conformance2_textures_webgl_canvas_tex_2d_srgb8_rgb_unsigned_byte * gpu_tests.webgl_conformance_integration_test.WebGLConformanceIntegrationTest.WebglConformance_conformance2_textures_webgl_canvas_tex_3d_r32f_red_float Also reproduces on V8's own WebGl bots with differing but related tests: https://build.chromium.org/p/client.v8.fyi/builders/Mac%20Release%20%28Intel%29/builds/989
,
Mar 27 2017
Duplicate of bug 705156 ?
,
Mar 27 2017
,
Mar 28 2017
Why does https://build.chromium.org/p/tryserver.chromium.mac/builders/mac_optional_gpu_tests_rel/builds/8166 say in step 72: TESTS FAILED; retries without patch disabled (disabled in recipes) Why are retries without patch disabled? Is that temporary or is that a missing feature for gpu tests? Without retries, any (unrelated) continuous failures on CI will block CQ, like it happened now.
,
Mar 28 2017
+phajdan, see last comment
,
Mar 28 2017
machenbach@: it's a missing feature because of limitations in the GPU tests' harness (more specifically, the isolated script test step). We disabled the retries without the patch because they would be tremendously expensive on this tryserver. Please feel free to file an RFE and assign it to me about this. But on the whole we aim to avoid situations where these tests would fail because of upstream breakage. This outage happened for a specific reason (because Skia wasn't running these tests in their autoroller) and that has been corrected.
,
Mar 29 2017
Thanks! |
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by kbr@chromium.org
, Mar 27 2017Components: Internals>Skia
Owner: kbr@chromium.org
Status: Assigned (was: Untriaged)