Closed Bug 1283579 Opened 8 years ago Closed 8 years ago

Intermittent dom/base/test/test_bug682592.html | Test timed out.

Categories

(Core :: Layout: Text and Fonts, defect)

Product:
Core
Component:
Layout: Text and Fonts
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED FIXED
Milestone:
mozilla50
Tracking Flags:
Tracking Status
firefox49 --- fixed
firefox50 --- fixed

People

(Reporter: intermittent-bug-filer, Assigned: xidorn)

References

Details

(Keywords: intermittent-failure)

Attachments

(2 files)

Bug 1283579 - Use data schema rather than outside document for test of bug 682592 to make it faster.
58 bytes, text/x-review-board-request
jfkthame
: review+
Details
distribution of load times
25.47 KB, image/png
Details 
Xidorn, can you please take a look at this new intermittent? It started with a low frequency but happens now very often (doesn't need to mean anything). Oldest occurrence I found is https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&revision=cbf85303aeb2cea6326574a39db03dad860b578f / https://treeherder.mozilla.org/logviewer.html#?job_id=30992369&repo=mozilla-inbound Thank you.
Component: DOM → Layout: Text
Flags: needinfo?(xidorn+moz)
That test is actually 136 reftests, which could take lots of time. I can optimize that test a bit, but if that doesn't solve the problem, we may just increase the number in requestLongerTimeout.
Flags: needinfo?(xidorn+moz)
See Also: → 703773
We'll see whether it works: https://treeherder.mozilla.org/#/jobs?repo=try&revision=00d69723f889

So according to my profiling (using the Performance API), the majority of the time in this test is taken in reloading for having a fresh document to test. That part takes over three quarters of the total time in the test. (Taking snapshots and comparing them take ~10% of the total time, and other profiled parts take <5% in total, and I don't know where is the remaining ~10% of time from.)

With my proposed patch, the time to run this test is cut down by 40% for me locally, which should make things significantly better, and shouldn't affect what the test is checking.
Assignee: nobody → xidorn+moz
Hmmm, that patch doesn't work. It means, unlike on Windows, the majority of time is not spent on loading for Android on this test. Then we need some further investigating.
Some data from this try push: https://treeherder.mozilla.org/#/jobs?repo=try&revision=3afcc7674bda
> Time summary:
>     Time for initial:  7810.8150000000005
>     Time for loading:  530476.72
>     Time for test:     11286.480000000403
>     Time for snapshot: 24172.755000000026

So the majority of time is still spent on loading... How can loading be so slow? Each reload could take 4s in average even without any network request?

Then I think the only way to fix is to increase the number in requestLongerTimeout. I guess the optimization is still worth landing given we've been here.
Comment on attachment 8767547 [details]
Bug 1283579 - Use data schema rather than outside document for test of bug 682592 to make it faster.

https://reviewboard.mozilla.org/r/61998/#review58790

This seems helpful, but given that it doesn't actually fix the problem on android, we shouldn't resolve the bug for now.
Attachment #8767547 - Flags: review?(jfkthame) → review+
(In reply to Xidorn Quan [:xidorn] (UTC+10) from comment #8)
> Some data from this try push:
> https://treeherder.mozilla.org/#/jobs?repo=try&revision=3afcc7674bda
> > Time summary:
> >     Time for initial:  7810.8150000000005
> >     Time for loading:  530476.72
> >     Time for test:     11286.480000000403
> >     Time for snapshot: 24172.755000000026
> 
> So the majority of time is still spent on loading... How can loading be so
> slow? Each reload could take 4s in average even without any network request?

That seems .... terrible.

I wonder what the distribution of load times is like. One possibility is that most of them are reasonable, and then there's one (or a handful of cases) where a painfully slow GC steps in, or something like that, and the load ends up blocked for ages.
So it seems most of loads end in ~2s, and reguarly there are some loads take ~5s and some take ~25s.
Comment on attachment 8767547 [details]
Bug 1283579 - Use data schema rather than outside document for test of bug 682592 to make it faster.

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/61998/diff/1-2/
I suppose you're fine with me increasing the number for requestLongerTimeout.
Pushed by xquan@mozilla.com:
https://hg.mozilla.org/integration/autoland/rev/045e13e67ed4
Use data schema rather than outside document for test of bug 682592 to make it faster. r=jfkthame
https://hg.mozilla.org/mozilla-central/rev/045e13e67ed4
Status: NEW → RESOLVED
Closed: 8 years ago
status-firefox50: --- → fixed
Resolution: --- → FIXED
Target Milestone: --- → mozilla50
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: