Closed Bug 858055 Opened 11 years ago Closed 11 years ago

Intermittent test_bug669671.html | Load count (should be cached) - got 10, expected 8

Categories

(Core :: DOM: Navigation, defect)

Product:
Core
Component:
DOM: Navigation
Platform:
ARM
Android
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED FIXED
Milestone:
mozilla24

People

(Reporter: RyanVM, Assigned: mcmanus)

Details

(Keywords: intermittent-failure)

Attachments

(1 file)

disable cache_lock_stalled detector for tests that rely on cache
949 bytes, patch
jmaher
: review+
Details | Diff | Splinter Review 
https://tbpl.mozilla.org/php/getParsedLog.php?id=21427591&tree=Mozilla-Central

Android Tegra 250 mozilla-central opt test mochitest-3 on 2013-04-03 22:09:04 PDT for push c232bec6974d
slave: tegra-078

17170 INFO TEST-START | /tests/docshell/test/test_bug669671.html
17171 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17172 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17173 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17174 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17175 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17176 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17177 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17178 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17179 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17180 INFO TEST-PASS | /tests/docshell/test/test_bug669671.html | Load count
17181 ERROR TEST-UNEXPECTED-FAIL | /tests/docshell/test/test_bug669671.html | Load count (should be cached) - got 10, expected 8
17182 INFO TEST-END | /tests/docshell/test/test_bug669671.html | finished in 4173ms
Perhaps this is related to bug 855762? I made some speculation in bug 855762, comment 135.
It appears that bug 868441, was checked in and nothing was built on inbound when it was.  The tests before this do not exhibit this problem, but the tests after it do.  

For reference, this specific problem failed prior to this test, but very rarely, now it seems to fail much more often.

Patrick, could you take a look at that patch and this docshell test case and come up with any reason why it would be causing this test case to fail quite often?
Flags: needinfo?(mcmanus)
The tegra's disk cache is responding so amazingly slowly that it is being autodisabled. (that's the point of that patch). The test counts the number of times something is cached - which is obviously influenced by this :)

The earlier fails in april were probably related to a similar piece of code that got backed out.

I can modify this test to greatly lower the sensitivity of the auto-disabling.
Flags: needinfo?(mcmanus)
this makes good sense, please submit a patch to the test case!
Patrick- you mentioned in comment 96 that you could put a patch together to fix this on android.  This bug is a top issue in orange factor.
Flags: needinfo?(mcmanus)
there is a patch at try now.
Flags: needinfo?(mcmanus)
setting the pref for just that test didn't work.. in retrospect that makes sense - anytime the cache is observed to be slow it can be disabled and dampened for a while.. so it is probably that earlier decision that is being noticed in this test.

Is there a pref file that covers all of mochi? (or at least mochi-3?)
for all mochitests:
http://mxr.mozilla.org/mozilla-central/source/testing/profiles/prefs_general.js

If we want to do prefs specific for android, we would need to hack the remote runner (similar to this):
http://mxr.mozilla.org/mozilla-central/source/testing/mochitest/runtestsremote.py#583
Assignee: nobody → mcmanus
Attachment #755029 - Flags: review?(jmaher)
Comment on attachment 755029 [details] [diff] [review]
disable cache_lock_stalled detector for tests that rely on cache

Review of attachment 755029 [details] [diff] [review]:
-----------------------------------------------------------------

this looks good.  Any concern with this on desktop?  If not, lets land it!
Attachment #755029 - Flags: review?(jmaher) → review+
(In reply to Joel Maher (:jmaher) from comment #149)
> Comment on attachment 755029 [details] [diff] [review]
> disable cache_lock_stalled detector for tests that rely on cache
> 
> Review of attachment 755029 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> this looks good.  Any concern with this on desktop?  If not, lets land it!

no concerns about desktop - this is the right behavior for the tests everywhere. (desktop just isn't slow enough in general to trigger the auto-disable logic so its effectively a nop there.. Although I think I saw at least 1 desktop orange.)

I'll push when inbound opens.
https://hg.mozilla.org/mozilla-central/rev/dbccec261411
Status: NEW → RESOLVED
Closed: 11 years ago
Flags: in-testsuite+
Resolution: --- → FIXED
Target Milestone: --- → mozilla24
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: