The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build/+/5b8ece0ca2534f87fe5527f6ba5ce2c2d9969381

commit 5b8ece0ca2534f87fe5527f6ba5ce2c2d9969381
Author: Nico Weber <thakis@chromium.org>
Date: Mon Apr 16 16:12:31 2018

Remove test_installer from win/asan bots again.

They don't pass there, see first bug below.

Bug: 833411, 827082 
Change-Id: Ic0acb2c9c67b7c3d6d8019f6467418f2fe31d555
Reviewed-on: https://chromium-review.googlesource.com/1014025
Reviewed-by: Hans Wennborg <hans@chromium.org>
Commit-Queue: Nico Weber <thakis@chromium.org>

[modify] https://crrev.com/5b8ece0ca2534f87fe5527f6ba5ce2c2d9969381/scripts/slave/recipe_modules/chromium_tests/chromium_clang.py

Assigning to mmeade@, who is actively working on the test these days.

I see these in the logs:

[0414/040923:test_installer.py(450)] Config found at C:\b\c\b\CrWinAsan\src\chrome\test\mini_installer\config\config.config
[0414/040923:test_installer.py(509)] Setting --force-clean
ChromeUserLevel: no_pv -> install_chrome_user -> chrome_user_installed_not_inuse -> test_chrome_with_chromedriver_user -> chrome_user_installed_not_inuse -> uninstall_chrome_user -> clean
 ... [0414/040924:test_installer.py(136)] Verifying state no_pv
[0414/040924:test_installer.py(103)] Beginning action install_chrome_user

command timed out: 2400 seconds without output, attempting to kill

So it looks like the install step hung. There isn't any output from the test after the install step. The timeout error came from the runner. 

Unfortunately the test itself can't grab logs when it is killed. Any ideas why it would hang after 40 minutes of testing? Is there a difference in terms of the intstaller with the asan runs?

I would like to leave this bug open as a tracking bug since it already has all of the info.

 crbug.com/33407  is about the installer growing too large for 7z to zip up next_version_mini_installer. This one has already gotten past that point, but it seems there is something on asan that is causing these issues. 

Greg noted that it looks like either the binary grew, or just started running. 

Does anyone know if this target was running on asan before Nico enabled it?

I don't think the two issues are the same. Is there a way to pull the built files off of somewhere? If not, you could try building locally with the same GN args to see if you can repro. From the generate_build_files step:

clang_use_chrome_plugins = false
enable_ipc_fuzzer = true
is_asan = true
is_clang = true
is_component_build = false
is_debug = false
llvm_force_head_revision = true
symbol_level = 2
target_cpu = "x86"
v8_enable_verify_heap = true