Findit identified the culprit r581558 with confidence 70.0% in the config "chromium.memory / Linux Chromium OS ASan LSan Tests (1)"
based on the flakiness trend:

https://findit-for-me.appspot.com/waterfall/flake?key=ag9zfmZpbmRpdC1mb3ItbWVy8gELEhdNYXN0ZXJGbGFrZUFuYWx5c2lzUm9vdCK7AWNocm9taXVtLm1lbW9yeS9MaW51eCBDaHJvbWl1bSBPUyBBU2FuIExTYW4gVGVzdHMgKDEpLzI4NTUyL3N5bmNfaW50ZWdyYXRpb25fdGVzdHMvVkhkdlEyeHBaVzUwVUhKbFptVnlaVzVqWlhOVGVXNWpWR1Z6ZEM1VGFXNW5iR1ZEYkdsbGJuUkZibUZpYkdWa1JXNWpjbmx3ZEdsdmJrSnZkR2hEYUdGdVoyVmtYMFV5UlZSbGMzUT0MCxITTWFzdGVyRmxha2VBbmFseXNpcxgBDA

If the culprit above is wrong, please file a bug using this link and hit submit:
https://bugs.chromium.org/p/chromium/issues/entry?status=Unconfirmed&labels=Pri-1,Test-Findit-Wrong&components=Tools%3ETest%3EFindit%3EFlakiness&summary=%5BFindit%5D%20Flake%20Analyzer%20-%20Wrong%20result%20for%20TwoClientPreferencesSyncTest.SingleClientEnabledEncryptionBothChanged_E2ETest&comment=Link%20to%20Analysis%3A%20https://findit-for-me.appspot.com/waterfall/flake?key=ag9zfmZpbmRpdC1mb3ItbWVy8gELEhdNYXN0ZXJGbGFrZUFuYWx5c2lzUm9vdCK7AWNocm9taXVtLm1lbW9yeS9MaW51eCBDaHJvbWl1bSBPUyBBU2FuIExTYW4gVGVzdHMgKDEpLzI4NTUyL3N5bmNfaW50ZWdyYXRpb25fdGVzdHMvVkhkdlEyeHBaVzUwVUhKbFptVnlaVzVqWlhOVGVXNWpWR1Z6ZEM1VGFXNW5iR1ZEYkdsbGJuUkZibUZpYkdWa1JXNWpjbmx3ZEdsdmJrSnZkR2hEYUdGdVoyVmtYMFV5UlZSbGMzUT0MCxITTWFzdGVyRmxha2VBbmFseXNpcxgBDA

Automatically posted by the findit-for-me app (https://goo.gl/Ot9f7N).

Could you check whether the CL above causing this and if so revert it?

Unfortunately, I can't evaluate.

+jkrcal@ as a triager.

+ the author of the CL.

Swapping owners so that the author of the CL is the owner.

Rouslan, could you please check your culprit CL?

Interesting! Let me take a look at what TwoClientPreferencesSyncTest.SingleClientEnabledEncryptionBothChanged_E2ETest does...

The patch seems unrelated, it's more likely a test flake.

While I tend to agree with mastiz@, keep in mind that the patch does add one preference "kAutofillJapanCityFieldMigrated" that is flipped from "false" to "true" in PersonalDataManager::MoveJapanCityToStreetAddress():

https://cs.chromium.org/chromium/src/components/autofill/core/browser/personal_data_manager.cc?rcl=ec978fdcd324d32e1dd1ccec03900084652ebe26&l=1497

This method runs from PersonalDataManager::ApplyAddressFixesAndCleanups(), which is called from PersonalDataManager::SyncStarted(syncer::AUTOFILL_PROFILE).

The boolean preference is there to prevent the migration code from running more than once. Perhaps we should set it to "false" in the sync tests to avoid the migration code from interfering with the tests.

We should add something like this in TwoClientPreferencesSyncTest before starting the sync service:

pref_service_->SetBoolean(prefs::kAutofillJapanCityFieldMigrated, true);

Do you think this would help?

Alternative solution is to make kAutofillJapanCityFieldMigrated non-syncable, which should not have too large of an impact on performance.

How does solution in #10 fix the problem? We still need to override the default value of the pref in the tests, don't we?

Anyway, do I understand it correctly that you plan to revert / elaborate on the problem?

> How does solution in #10 fix the problem?

I expect a non-syncable preference to not interfere with a "preferences sync test."

> Do I understand it correctly that you plan to revert / elaborate on the problem?

I prefer to avoid a revert, so I will submit a couple of speculative patches in hopes that FindIt will detect that the flakes have stopped.

Findit identified the culprit r581558 with confidence 70.0% in the config "chromium.chromiumos / linux-chromeos-dbg"
based on the flakiness trend:

https://findit-for-me.appspot.com/waterfall/flake?key=ag9zfmZpbmRpdC1mb3ItbWVy4gELEhdNYXN0ZXJGbGFrZUFuYWx5c2lzUm9vdCKrAWNocm9taXVtLmNocm9taXVtb3MvbGludXgtY2hyb21lb3MtZGJnLzcyMTAvc3luY19pbnRlZ3JhdGlvbl90ZXN0cy9WSGR2UTJ4cFpXNTBVSEpsWm1WeVpXNWpaWE5UZVc1alZHVnpkQzVUYVc1bmJHVkRiR2xsYm5SRmJtRmliR1ZrUlc1amNubHdkR2x2YmtKdmRHaERhR0Z1WjJWa1gwVXlSVlJsYzNRPQwLEhNNYXN0ZXJGbGFrZUFuYWx5c2lzGAEM

If the culprit above is wrong, please file a bug using this link and hit submit:
https://bugs.chromium.org/p/chromium/issues/entry?status=Unconfirmed&labels=Pri-1,Test-Findit-Wrong&components=Tools%3ETest%3EFindit%3EFlakiness&summary=%5BFindit%5D%20Flake%20Analyzer%20-%20Wrong%20result%20for%20TwoClientPreferencesSyncTest.SingleClientEnabledEncryptionBothChanged_E2ETest&comment=Link%20to%20Analysis%3A%20https://findit-for-me.appspot.com/waterfall/flake?key=ag9zfmZpbmRpdC1mb3ItbWVy4gELEhdNYXN0ZXJGbGFrZUFuYWx5c2lzUm9vdCKrAWNocm9taXVtLmNocm9taXVtb3MvbGludXgtY2hyb21lb3MtZGJnLzcyMTAvc3luY19pbnRlZ3JhdGlvbl90ZXN0cy9WSGR2UTJ4cFpXNTBVSEpsWm1WeVpXNWpaWE5UZVc1alZHVnpkQzVUYVc1bmJHVkRiR2xsYm5SRmJtRmliR1ZrUlc1amNubHdkR2x2YmtKdmRHaERhR0Z1WjJWa1gwVXlSVlJsYzNRPQwLEhNNYXN0ZXJGbGFrZUFuYWx5c2lzGAEM

Automatically posted by the findit-for-me app (https://goo.gl/Ot9f7N).

This seems to be being actively investigated -> taking out of sheriff queue.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0b370929d255471e3122bbd9ab0ecbf0cf310cb0

commit 0b370929d255471e3122bbd9ab0ecbf0cf310cb0
Author: Rouslan Solomakhin <rouslan@chromium.org>
Date: Mon Aug 13 16:19:47 2018

Make AutofillJapanCityFieldMigrated preference non-syncable

This is a speculative fix to flakiness in
TwoClientPreferencesSyncTest.SingleClientEnabledEncryptionBothChanged_E2ETest.
https://crrev.com/c/1152068 introduced AutofillJapanCityFieldMigrated as
a syncable preference that starts out as "false" by default, but then is
flipped to "true" upon the start of syncing of autofill profiles. The
speculation is that a non-syncable preference should not interfere with
a preferences sync test.

Bug:  872614 
Change-Id: I84cc91c8e11aeaba6a02ea178a5b38ebde97882a
Reviewed-on: https://chromium-review.googlesource.com/1169309
Reviewed-by: Roger McFarlane <rogerm@chromium.org>
Reviewed-by: Jan Krcal <jkrcal@chromium.org>
Commit-Queue: Rouslan Solomakhin <rouslan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#582605}
[modify] https://crrev.com/0b370929d255471e3122bbd9ab0ecbf0cf310cb0/components/autofill/core/common/autofill_prefs.cc

Did the test stop flaking?

According to https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=sync_integration_tests&tests=TwoClientPreferencesSyncTest.SingleClientEnabledEncryptionBothChanged it seems okay now.

The test was consistently crashing on some bots between Aug 25 and Aug 26 but this seems unrelated to your revert.