Are you sure that analysis is correct? https://codereview.chromium.org/2857643002/ disabled in-process symbolication for asan builds because in that setup we run an external symbolizer that's better at symbolizing. The bots should already pipe through asan_symbolize. It's possible that that part broke, but that'd be independent of https://codereview.chromium.org/2857643002/

I was also recently surprised by the removal of the symbolization, as I run all of my development with ASan builds. This means I need to now have an explicit step of copy/run symbolizer/paste to decode or pipe through the symbolizer. However, I already pipe output through some other tools that help out with output, which makes the command line even longer. I might be one of the rare cases, but wanted to voice that this change made my dev workflow much harder :(.

nasko: That sounds unrelated to what this bug is about. Feel free to say that on ttps://codereview.chromium.org/2857643002/ . The motivation behind that change was to a) _improve_ local dev by getting better stacks b) let devs do the same thing that bots already do (...which used to be "pipe stuff through asan_symbolize", which is why I'm fairly certain that the lost symbols on the asan bots aren't due to my change)

thakis@: I see, I didn't realize the bots should already do this. Hopefully infra team can triage this into the right component and we can figure out what's going on here.

(I've noticed this issue for crashes in layout test runs as well... if I see an instance, I'll post it on here)

on behalf of infra, we have no component for this bug, other than general Infra>Client>Chrome, for which we have no owners.

Somebody who understands what's this about has to either
 * run recipe locally and try to repro, then fix problem
 * ssh to bot, and try to repro it there.

Where are the docs for running the recipe locally?

You don't need to run a recipe locally. Just do

out/gn/base_unittests | tools/valgrind/asan/asan_symbolize.py

(You might also need to export ASAN_SYMBOLIZER_PATH=/path/to/chromium/src/third_party/llvm-build/Release+Asserts/bin/llvm-symbolizer)

c#9: right, but that's only for local runs? This bug is about making the test autofiltered on the bots (which it currently isn't: see https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Ftryserver.chromium.linux%2Flinux_chromium_asan_rel_ng%2F399601%2F%2B%2Frecipes%2Fsteps%2Fcomponents_unittests__with_patch_%2F0%2Flogs%2FFaviconHandlerTest.UpdateSameIconURLsWhileDatabaseLookupShouldBeNoop%2F0 as another example)

Fwiw here's a symbolized stack on the same bot: https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Ftryserver.chromium.linux%2Flinux_chromium_asan_rel_ng%2F402239%2F%2B%2Frecipes%2Fsteps%2Fmedia_unittests__with_patch_%2F0%2Flogs%2FCdmAdapter__x2f_AesDecryptorTest.LatestKeyUsed__x2f_0%2F0

This shows that symbolizing on the bots basically works, and there's just some config issue somewhere independent of my cl.

https://codereview.chromium.org/2961603003 looks pretty related. Maybe we need that for more binaries.

I'm not sure if this issue belongs to I>C>C, but it sounds important enough that someone should at least keep track of it. I'm adding it to chrome-client-infra-backlog hotlist for tracking and possibly later re-triage. +jbudorick@ FYI.

This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

 Issue 781100  has been merged into this issue.