New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.
Starred by 3 users

Issue metadata

Status: Fixed
Closed: Dec 13
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 2
Type: Bug

Sign in to add a comment

Issue 789129: Disabled flaky SingleClientSessionsSyncTest.CookieJarMismatch test

Reported by, Nov 28 2017 Project Member

Issue description

It's flaking again, assigning to owner who previously dealt with this.


Comment 1 by, Nov 28 2017

Project Member
The following revision refers to this bug:

commit 66f71311862cd3786f54348db6d585963cc53359
Author: Henrik Boström <>
Date: Tue Nov 28 13:33:38 2017

Disabled flaky SingleClientSessionsSyncTest.CookieJarMismatch test.,

No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug:  789129 
Change-Id: I46ac7f186e52770b09a39be276ee5959ed56a1c5
Reviewed-by: Henrik Boström <>
Commit-Queue: Henrik Boström <>
Cr-Commit-Position: refs/heads/master@{#519668}

Comment 2 by, Nov 29 2017

Status: Started (was: Assigned)

Comment 3 by, Dec 5 2017

Project Member
The following revision refers to this bug:

commit 34683f17ddf749befcc482135f68f397f8809ec0
Author: Pavel Yatsuk <>
Date: Tue Dec 05 01:43:03 2017

[Sync] Verify that correct value of CookieJarMismatch is passed to SyncCycleContext

SingleClientSessionsSyncTest.CookieJarMismatch test is flaky. It fails on the
step when signed in account gets included in the list of accounts in cookie jar.
In this case value of cookie_jar_mismatch should turn to false.

I wasn't able to reproduce this bug, nor I was able to find scenario how this
test could fail.

I'm adding a check that ensures correct value of cookie_jar_mismatch is passed
to sync engine. The next time this test fails it will be clear if failure caused
by incorrect value or the value wasn't delivered to the server.
BUG= 789129 

Change-Id: I508e5d37083e98b2c02e1bf3a35917dad3a9645f
Commit-Queue: Pavel Yatsuk <>
Reviewed-by: Patrick Noland <>
Cr-Commit-Position: refs/heads/master@{#521583}

Comment 4 by, Dec 5 2017

Labels: -Pri-1 Pri-2
I've added some more checks to the test to isolate the failure the next time it happens. I'm lowering priority of this bug. Will investigate failures when they occur.

Comment 5 by, Dec 5 2017

 Issue 786867  has been merged into this issue.

Comment 6 by, Dec 8 2017

Labels: -Pri-2 OS-Linux Pri-1
a few more flakes:

Also this is still causing bots on the CQ to fail, potentially blocking devs from landing CLs.  Hence raising the priority again.

Comment 7 by, Dec 12 2017

I'm going to disable the test again based on "I'm lowering priority of this bug. Will investigate failures when they occur."

There are new failures to investigate in c#6, and probably more since then.

Comment 8 by, Dec 13 2017

Had the wrong bug on the CL so bugdroid didn't pick it up.  CL has landed as

Comment 9 by, Dec 13 2017

Project Member
The following revision refers to this bug:

commit 9f4ce7cb311351aecae213234812df8d7f1b9b4d
Author: Pavel Yatsuk <>
Date: Wed Dec 13 18:10:36 2017

[Sync] Add logging to diagnose SingleClientSessionsSyncTest.CookieJarMismatch flakines

I'm logging contents of FakeServer before the call to UpdateCookieJarAccountsAndWait
to ensure that later call to NavigateTab will indeed trigger commit.

I'm adding logging of last commit message to ensure that navigation to kURL2
triggered sync cycle.
BUG= 789129 

Change-Id: I7adb3750f5aa2c5e4611fc32b37df461dc744d49
Reviewed-by: Patrick Noland <>
Commit-Queue: Pavel Yatsuk <>
Cr-Commit-Position: refs/heads/master@{#523818}

Comment 10 by, Jan 16 2018

Labels: -Pri-1 Pri-2

Comment 11 by, Jan 17 2018

Labels: SyncHandoff2018
Owner: ----
Status: Available (was: Started)

Comment 12 by, Apr 27 2018

Components: Tests>Flaky
Labels: SingleClientSessionsSyncTest.CookieJarMismatch Test-Disabled

Comment 13 by, Jun 7 2018

Labels: -Pri-2 Pri-3
Lowering prio: I believe Dice will make this functionality (detecting cookie jar mismatch) and thus the test obsolete.

Comment 14 by, Sep 6

Labels: -Pri-3 Pri-2
Looks like it won't be so easy: Even Chrome versions that have DICE still report non-trivial numbers of mismatches to UMA. (It's unclear whether there are real mismatches, or there's something wrong with how we detect them.) The cookie_jar_mismatch flag is also sent to the server.
So we probably can't get rid of this for the foreseeable future after all :(

Comment 15 by, Sep 6


Comment 16 by, Dec 10

Labels: FixitSync
Status: Assigned (was: Available)

Comment 17 by, Dec 10

Status: Started (was: Assigned)

Comment 18 by, Dec 10

Project Member
The following revision refers to this bug:

commit b4fc383409e81c48285ef1de7c08a6fdc1311a4f
Author: Mikel Astiz <>
Date: Mon Dec 10 16:23:05 2018

Reenable SingleClientSessionsSyncTest.CookieJarMismatch

The test has been historically flaky but many things have changed since
the last reported flakes (in particular, migration of SESSIONS to USS)
so chances are that we no longer repro the flakes.

We also do some related code cleanup.

Bug:  789129 
Change-Id: I7f70d8d0d1c96232ffe07197e8e1982759e8af02
Commit-Queue: Mikel Astiz <>
Reviewed-by: Marc Treib <>
Cr-Commit-Position: refs/heads/master@{#615138}

Comment 19 by, Dec 11

Here's a recent flake (

../../chrome/browser/sync/test/integration/ Failure
Value of: message.commit().config_params().cookie_jar_mismatch()
  Actual: true
Expected: false
   "client_status": {
      "hierarchy_conflict_detected": false
   "commit": {
      "cache_guid": "wzo0ggZ7J/KpLvl+/P0E7Q==",
      "config_params": {
         "cookie_jar_mismatch": true,
         "enabled_type_ids": [ "32904", "37702", "45873", "63951", "31729", "306270", "330441", "41210", "40781", "48119", "88610", "50119", "48364", "103656", "96159", "150251", "170540", "182019", "181534", "154522", "163425", "455206", "556014", "47745", "161496" ],
         "tabs_datatype_enabled": true
      "entries": [ {
         "specifics": {
            "user_event": {
               "event_time_usec": "13188966756772762",
               "language_detection_event": {
                  "detected_languages": [ {
                     "is_reliable": false,
                     "language_code": "und"
                  } ]
               "navigation_id": "13188966756622378",
               "session_id": "2817817161013833761"
      } ]
   "debug_info": {
      "cryptographer_has_pending_keys": false,
      "cryptographer_ready": true,
      "events_dropped": false
   "protocol_version": "52",
   "share": "",
   "store_birthday": "0"

Comment 20 by, Dec 11

Project Member
The following revision refers to this bug:

commit 9b8a1977e84b471e42a6b6798dc2a607dfb28e91
Author: Mikel Astiz <>
Date: Tue Dec 11 12:54:45 2018

Fix flakiness in SingleClientSessionsSyncTest.CookieJarMismatch

Unless something is done about it, there is a race between signin code
trying to fetch "real" cookiejar information using GaiaAuthFetcher, and
the test faking equivalent information and injecting it via

Usually, signin logic is fast enough to avoid test failures, but
occasionally the fetching takes long and an empty list of accounts is
injected at a last stage (after navigations to kURL2), interfering with
test expectations.

Bug:  789129 
Change-Id: I95b0f691bd047216309888d345c68e1eb26ab869
Reviewed-by: Marc Treib <>
Commit-Queue: Mikel Astiz <>
Cr-Commit-Position: refs/heads/master@{#615504}

Comment 21 by, Dec 13

Status: Fixed (was: Started)
All green except for Win10 Tests x64 (dbg) where was have issues with most sync tests in general.

Comment 22 by, Jan 25

Labels: Test-Flaky

Comment 23 by, Jan 25

Labels: Test-Flaky

Sign in to add a comment