951181 - [meta] "Automation Error: Missing end of test marker (process crashed?)" after some robocop runs

Status: RESOLVED WONTFIX

(Firefox for Android Graveyard :: Testing, defect)

Description

[Ryan VanderMeulen [:RyanVM]]

Happens after the same tests on all Android platforms we test.

https://tbpl.mozilla.org/php/getParsedLog.php?id=32073208&tree=Mozilla-Central

Android 2.2 Armv6 Tegra mozilla-central opt test robocop-1 on 2013-12-16 22:32:18 PST for push b1e5ade62913
slave: tegra-238

351 INFO TEST-PASS | testBrowserProvider - TestExpireHistory | Expected number of inserts matches - 3000 should equal 3000
INFO | automation.py | Application ran for: 0:05:01.264851
INFO | zombiecheck | Reading PID log: /tmp/tmpsRGFgVpidlog
/data/anr/traces.txt not found
WARNING | leakcheck | refcount logging is off, so leaks can't be detected!
runtests.py | Running tests: end.
Mochi-Remote WARNING | cleaning up pidfile '/builds/tegra-238/test/../runtestsremote.pid' was unsuccessful from the test harness
Mochi-Remote ERROR   | Automation Error: Missing end of test marker (process crashed?)

38 INFO TEST-PASS | testDistribution | clean up distribution files - deleted /data/data/org.mozilla.fennec/distribution
Clearing distribution pref.
INFO | automation.py | Application ran for: 0:00:15.103412
INFO | zombiecheck | Reading PID log: /tmp/tmpKyH0GOpidlog
/data/anr/traces.txt not found
WARNING | leakcheck | refcount logging is off, so leaks can't be detected!
runtests.py | Running tests: end.
Mochi-Remote WARNING | cleaning up pidfile '/builds/tegra-238/test/../runtestsremote.pid' was unsuccessful from the test harness
Mochi-Remote ERROR   | Automation Error: Missing end of test marker (process crashed?)


https://tbpl.mozilla.org/php/getParsedLog.php?id=32073071&tree=Mozilla-Central

Android 2.2 Armv6 Tegra mozilla-central opt test robocop-2 on 2013-12-16 22:32:17 PST for push b1e5ade62913
slave: tegra-195

63 INFO TEST-PASS | testSystemPages | The correct number of tabs are opened - 4 should equal 4
INFO | automation.py | Application ran for: 0:05:01.320126
INFO | zombiecheck | Reading PID log: /tmp/tmpnltLyfpidlog
/data/anr/traces.txt not found
WARNING | leakcheck | refcount logging is off, so leaks can't be detected!
runtests.py | Running tests: end.
Mochi-Remote WARNING | cleaning up pidfile '/builds/tegra-195/test/../runtestsremote.pid' was unsuccessful from the test harness
Mochi-Remote ERROR   | Automation Error: Missing end of test marker (process crashed?)

Comment 1

[Geoff Brown [:gbrown] (pto Apr 11-18)]

:margaret -- Did I see you working on this somewhere?

Comment 2

[:Margaret Leibovic]

Did this just start after bug 949208 landed? That patch fixed an NPE at the end of ContentProviderTests, but maybe there's still something else going wrong?

Comment 3

[Ryan VanderMeulen [:RyanVM]]

Sorry, I don't know exactly when this started. I just went on a filing spree today for various logspam issues that are turning up.

Comment 4

[Michael Comella (:mcomella) [NI reported issues only: ex-Mozilla]]

These failures seem individually based:

---
testBrowserProvider didn't seem to finish - it's missing some asserts:

350 INFO TEST-PASS | testBrowserProvider - TestExpireHistory | All thumbnails were deleted - 0 should equal 0
351 INFO TEST-PASS | testBrowserProvider - TestExpireHistory | Expected number of inserts matches - 3000 should equal 3000
INFO | automation.py | Application ran for: 0:05:01.264851

vs. a success:

351 INFO TEST-PASS | testBrowserProvider - TestExpireHistory | Expected number of inserts matches - 3000 should equal 3000
352 INFO TEST-PASS | testBrowserProvider - TestExpireHistory | 3000 history entries found - 3000 should equal 3000
353 INFO TEST-PASS | testBrowserProvider - TestExpireHistory | Expected number of inserts matches - 3000 should equal 3000
354 INFO TEST-PASS | testBrowserProvider - TestExpireHistory | 3000 thumbnails entries found - 3000 should equal 3000
355 INFO TEST-PASS | testBrowserProvider - TestExpireHistory | 500 history entries found - 500 should equal 500
356 INFO TEST-PASS | testBrowserProvider - TestExpireHistory | 15 thumbnails found - 15 should equal 15
357 INFO TEST-START | testBrowserProvider - TestBrowserProviderNotifications
...

The first two lines of /dev/log/main that repeat over and over again is suspicious. They are:

12-16 22:55:02.059 W/GeckoProfile( 4212): requested profile directory missing: null
12-16 22:55:02.059 W/GeckoBrowserProvider( 4212): Tried to insert an empty or non-byte-array image. Ignoring.

Otherwise, I'm not sure why this is failing.

---
testDistribution crashes:

12-16 22:59:36.789 I/TestRunner( 5280): java.lang.reflect.InvocationTargetException
12-16 22:59:36.789 I/TestRunner( 5280): 	at org.mozilla.gecko.db.BrowserProvider.getDatabasePath(BrowserProvider.java:1959)
12-16 22:59:36.789 I/TestRunner( 5280): 	at java.lang.reflect.Method.invokeNative(Native Method)
12-16 22:59:36.789 I/TestRunner( 5280): 	at java.lang.reflect.Method.invoke(Method.java:521)
12-16 22:59:36.789 I/TestRunner( 5280): 	at org.mozilla.gecko.tests.ContentProviderTest.tearDown(ContentProviderTest.java:257)

However, this reflection has since been removed [1].

---
testSystemPages seems to crash every time (I checked the 3 most recent m-c TBPL logs). The bad news is that means this test is never actually completing. I'm suspicious of:

12-16 23:03:19.035 I/ActivityManager( 1019): Displayed activity org.mozilla.fennec/org.mozilla.gecko.preferences.GeckoPreferences: 401 ms (total 401 ms)

which occurs less than a second before the last assertion is logged. I thought testing multiple activities in Android is really difficult, so maybe opening this Activity is killing the Activity under testing, ending the test early.

[1]: https://mxr.mozilla.org/mozilla-central/source/mobile/android/base/tests/ContentProviderTest.java#257

Comment 5

[Michael Comella (:mcomella) [NI reported issues only: ex-Mozilla]]

> testSystemPages seems to crash every time 

Actually, this appears to only be on Android 2.2. Interestingly, 4.0 seems fine.

Comment 6

[Michael Comella (:mcomella) [NI reported issues only: ex-Mozilla]]

Renamed this to a meta bug and filed individual bugs as blocking.

Comment 7

[Geoff Brown [:gbrown] (pto Apr 11-18)]

Those cases that happen after a run time of about 5 minutes are caused by bug 963838. I think that will prove to be the majority of the problem, but it may not cover all cases.

Comment 8

[Geoff Brown [:gbrown] (pto Apr 11-18)]

Mass closing Firefox for Android :: Testing bugs with no progress in 2017.

If this bug is important to you, please re-open.