Closed Bug 1000883 Opened 10 years ago Closed 10 years ago

**Only use if there is nothing else in the failure summary** Intermittent Emulator Tests IOError: Connection to Marionette server is lost. Check gecko.log (desktop firefox) or logcat (b2g) for errors.

Categories

(Testing :: Marionette Client and Harness, defect, P1)

Product:
Testing
Component:
Marionette Client and Harness
Platform:
ARM
Gonk (Firefox OS)
Type:
defect

Tracking

(Not tracked)

Status:
RESOLVED INCOMPLETE

People

(Reporter: cbook, Unassigned)

References

Details

(Keywords: intermittent-failure, pi-marionette-intermittent)

Attachments

(2 files)

Remove network hack from b2g reftests
1.52 KB, patch
automatedtester
: review+
Details | Diff | Splinter Review
Check for crashes on IOError in marionette
961 bytes, patch
automatedtester
: review+
Details | Diff | Splinter Review 
b2g_emulator_vm b2g-inbound opt test mochitest-5 on 2014-04-24 03:04:58 PDT for push f02007563885

slave: tst-linux64-spot-802

https://tbpl.mozilla.org/php/getParsedLog.php?id=38389995&tree=B2g-Inbound

IOError: Connection to Marionette server is lost. Check gecko.log (desktop firefox) or logcat (b2g) for errors.
Return code: 1
# TBPL FAILURE #
The mochitest suite: mochitest-5 ran with return status: FAILURE
I'm getting this error when I run tests on the emulator locally via mochitest-remote. Example consistent failure: ./mach mochitest-remote dom/events/test/test_bug432698.html
setting the priority since this is growing to be a monster
Priority: -- → P1
The failure seems to happen at this point every time:
http://mxr.mozilla.org/mozilla-central/source/build/mobile/b2gautomation.py#253

This seems like a mochitest specific hack (and the comment references a really old bug), I wonder if we can just remove it for reftest.. let's see what happens:
https://tbpl.mozilla.org/?tree=Try&rev=5ea1caadfd39
This should only affect b2g emulator reftests as only they still use b2gautomation.py. This means the bug will still affect mochitests.. but at least those seem way less frequent than the reftest ones.

The mochitest failures actually seem different, and have a crash stack attached. Maybe we should file a new bug for them.
Attachment #8419538 - Flags: review?(dburns)
Comment on attachment 8419538 [details] [diff] [review]
Remove network hack from b2g reftests

ship it
Attachment #8419538 - Flags: review?(dburns) → review+
https://hg.mozilla.org/integration/b2g-inbound/rev/33f2a1fa5663

Leaving open for now, since mochitests are still affected. We should file a separate bug to investigate the crash signature they have (if someone hasn't already).
Whiteboard: [leave-open]
The mochitest failures have the same crash signature as bug 931444.
Depends on: 931444
Ok, so this is still happening on reftest, it must have been coincidental timing that it was happening right at that "mochitest network hack" each time.

Something is intermittently crashing gecko it seems. Reftests don't have a crash stack, I discovered this is because marionette_transport now raises IOError instead of something derived from MarionetteException:
http://mxr.mozilla.org/mozilla-central/source/testing/marionette/transport/marionette_transport/transport.py#55

I believe this patch should give us crash stacks for reftest as well as mochitest.
Attachment #8420092 - Flags: review?(dburns)
Attachment #8420092 - Flags: review?(dburns) → review+
https://hg.mozilla.org/integration/b2g-inbound/rev/1509e459fd4f

Please note, I think I've done more or less everything I can about this bug (pending whether we get reftest crash stacks from this push). I think it is now up to devs to debug why this is crashing. I'll still gladly help out, just don't want to give the impression that I have everything under control.
I filed bug 1008357 for the js::ShapeTable crash. From the comments here, it appears to have started around 7-May.
Many of the recent starrings in this bug are mis-stars of crashes that deserve separate bugs.

This bug should only be starred if the failure excerpt does not contain anything else (eg a crash)

ie unless it's like comment 153 then don't star against this:

(In reply to TBPL Robot from comment #153)
> Tomcat
> https://tbpl.mozilla.org/php/getParsedLog.php?id=39730496&tree=Mozilla-
> Central
> b2g_emulator_vm mozilla-central opt test mochitest-6 on 2014-05-15 05:54:51
> revision: e4843f4f08a7
> slave: tst-linux64-spot-847
> 
> IOError: Connection to Marionette server is lost. Check gecko.log (desktop
> firefox) or logcat (b2g) for errors.
> Mochitest ERROR | Automation Error: Received unexpected exception while
> running application
> Return code: 1
> # TBPL FAILURE #
> The mochitest suite: mochitest-6 ran with return status: FAILURE
Summary: Intermittent Emulator Tests IOError: Connection to Marionette server is lost. Check gecko.log (desktop firefox) or logcat (b2g) for errors. → **Only use if there is nothing else in the failure summary** Intermittent Emulator Tests IOError: Connection to Marionette server is lost. Check gecko.log (desktop firefox) or logcat (b2g) for errors.
Given it has been 3-4 weeks since the last no-crash-stack instance in comment 153, I'm going to close this as incomplete for now. If more occur, then we need to look at why the comment 134 landing isn't catching all cases (comment 153 was at startup if I read the log correctly, so perhaps the crash reporter hadn't yet initialised).
URL: IOError: Connection to Marionette ser...
OS: Mac OS X → Gonk (Firefox OS)
Hardware: x86 → ARM
Whiteboard: [leave-open]
Version: unspecified → Trunk
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → INCOMPLETE
Product: Testing → Remote Protocol
Moving bug to Testing::Marionette Client and Harness component per bug 1815831.
Component: Marionette → Marionette Client and Harness
Product: Remote Protocol → Testing
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: