1672238 - Perma [Fission] TEST-UNEXPECTED-TIMEOUT | http://mochi.test:8888/tests/uriloader/exthandler/tests/mochitest/test_nullCharFile.xhtml | application timed out after 370 seconds with no output

Status: RESOLVED FIXED

(Firefox :: File Handling, defect, P2)

Description

[Treeherder Bug Filer]

Filed by: rmaries [at] mozilla.com
Parsed log: https://treeherder.mozilla.org/logviewer.html#?job_id=319145705&repo=mozilla-central
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/HMU7OshjTZ29px9o5UaBng/runs/0/artifacts/public/logs/live_backing.log

---

[task 2020-10-20T15:56:49.025Z] 15:56:49     INFO - TEST-START | http://mochi.test:8888/tests/uriloader/exthandler/tests/mochitest/test_nullCharFile.xhtml
[task 2020-10-20T16:13:16.493Z] 16:13:16     INFO - add_task | Entering test 
[task 2020-10-20T16:13:16.493Z] 16:13:16     INFO - Buffered messages finished
[task 2020-10-20T16:13:16.494Z] 16:13:16    ERROR - TEST-UNEXPECTED-TIMEOUT | http://mochi.test:8888/tests/uriloader/exthandler/tests/mochitest/test_nullCharFile.xhtml | application timed out after 370 seconds with no output
[task 2020-10-20T16:13:16.494Z] 16:13:16    ERROR - Force-terminating active process(es).
[task 2020-10-20T16:13:16.494Z] 16:13:16     INFO - Determining child pids from psutil...
[task 2020-10-20T16:13:16.522Z] 16:13:16     INFO - [20632, 20659, 20672, 20580, 20556, 20763, 20703, 20791, 20729]
[task 2020-10-20T16:13:16.525Z] 16:13:16     INFO - ==> process 20412 launched child process 20432
[task 2020-10-20T16:13:16.527Z] 16:13:16     INFO - ==> process 20412 launched child process 20556
[task 2020-10-20T16:13:16.529Z] 16:13:16     INFO - ==> process 20412 launched child process 20580
[task 2020-10-20T16:13:16.530Z] 16:13:16     INFO - ==> process 20412 launched child process 20632
[task 2020-10-20T16:13:16.531Z] 16:13:16     INFO - ==> process 20412 launched child process 20659
[task 2020-10-20T16:13:16.533Z] 16:13:16     INFO - ==> process 20412 launched child process 20672
[task 2020-10-20T16:13:16.534Z] 16:13:16     INFO - ==> process 20412 launched child process 20703
[task 2020-10-20T16:13:16.535Z] 16:13:16     INFO - ==> process 20412 launched child process 20729
[task 2020-10-20T16:13:16.535Z] 16:13:16     INFO - ==> process 20412 launched child process 20763
[task 2020-10-20T16:13:16.535Z] 16:13:16     INFO - ==> process 20412 launched child process 20791
[task 2020-10-20T16:13:16.535Z] 16:13:16     INFO - Found child pids: set([20672, 20580, 20556, 20432, 20659, 20791, 20632, 20729, 20763, 20703])
[task 2020-10-20T16:13:16.536Z] 16:13:16     INFO - Failed to get child procs
[task 2020-10-20T16:13:16.536Z] 16:13:16     INFO - Killing process: 20672
[task 2020-10-20T16:13:16.537Z] 16:13:16     INFO - TEST-INFO | started process screentopng
[task 2020-10-20T16:13:16.689Z] 16:13:16     INFO - TEST-INFO | screentopng: exit 0```

Comment 1

[Intermittent Failures Robot]

19 failures in 4230 pushes (0.004 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• mozilla-central: 16
• try: 3

Platform breakdown:

• linux1804-64: 19

For more details, see:
https://treeherder.mozilla.org/intermittent-failures.html#/bugdetails?bug=1672238&startday=2020-10-19&endday=2020-10-25&tree=all

Comment 2

[Marco Bonardo [:mak]]

This is apparenly only Linux debug, and what I see in the log is:

###!!! ASSERTION: AsyncWait failed: 'Error', file /builds/worker/checkouts/gecko/netwerk/base/nsInputStreamPump.cpp:128
WARNING: Blob data was not retrieved!: file /builds/worker/checkouts/gecko/dom/file/uri/BlobURLInputStream.cpp:437
#01: nsInputStreamPump::Resume() [netwerk/base/nsInputStreamPump.cpp:243]
#02: mozilla::net::DocumentLoadListener::FinishReplacementChannelSetup(nsresult) [netwerk/ipc/DocumentLoadListener.cpp:1078]
#03: mozilla::net::DocumentLoadListener::RedirectToRealChannelFinished(nsresult) [netwerk/ipc/DocumentLoadListener.cpp:1022]
...

I'm not sure what's up, I'll try reproducing on a Linux VM.

Comment 3

[Marco Bonardo [:mak]]

Attachment: Bug 1672238 - Fix test_nullCharFile.xhtml on fission x-origin. r=Gijs

Comment 4

[Marco Bonardo [:mak]]

My patch changes the Blob to a data uri, that for this test doesn't really matter, though we would not want to hide possible bugs here. I don't know much about why a Blob in x-orig would cause the network code to bail out, my theory is related to the Blob's origin, but I don't know the network and x-orig details well enough to go further.
Do you want me to file a follow-up bug to investigate this, or is it expected?

Comment 5

[Alexandru Michis [:malexandru]]

Marco, is your patch ok to be landed?

Comment 6

[Marco Bonardo [:mak]]

Yes, we can investigate the blob thing apart, it is not necessary for the test itself.

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/e301b90c2b0e
Fix test_nullCharFile.xhtml on fission x-origin. r=Gijs

Comment 8

[Atila Butkovits]

https://hg.mozilla.org/mozilla-central/rev/e301b90c2b0e

Comment 9

[Intermittent Failures Robot]

5 failures in 4275 pushes (0.001 failures/push) were associated with this bug in the last 7 days.

Repository breakdown:

• mozilla-central: 5

Platform breakdown:

• linux1804-64: 5

For more details, see:
https://treeherder.mozilla.org/intermittent-failures.html#/bugdetails?bug=1672238&startday=2020-10-26&endday=2020-11-01&tree=all

Comment 10

[Alexandru Ionescu (needinfo me) [:alexandrui]]

== Change summary for alert #27538 (as of Fri, 06 Nov 2020 07:47:05 GMT) ==

Improvements:

Ratio	Suite	Test	Platform	Options	Absolute values (old vs new)
9%	pinterest	Similarity	linux64-shippable-qr	cold webrender	0.86 -> 0.94
8%	pinterest	Similarity2D	linux64-shippable-qr	cold webrender	0.92 -> 0.99

For up to date results, see: https://treeherder.mozilla.org/perfherder/alerts?id=27538

Comment 11

[Kris Maglione [:kmag]]

The problem here is that when running in xorigin mode, the parent frame and the blob URI have different agent cluster IDs. And, since the load is happening in the frame, we compare the agent cluster ID of the parent frame to the one of the blob here and bail out.

I think that, for the purpose of this test, that doesn't matter, so the change is fine. But it also seems wrong. I'd expect an <a download> with a blob created by the page itself to work, regardless of the agent cluster of the parent frame. So that probably deserves a separate bug. Can you file one, please?

Comment 12

[Kris Maglione [:kmag]]

(In reply to Alexandru Ionescu (needinfo me) [:alexandrui] from comment #10)

== Change summary for alert #27538 (as of Fri, 06 Nov 2020 07:47:05 GMT) ==

Improvements:

Ratio	Suite	Test	Platform	Options	Absolute values (old vs new)
9%	pinterest	Similarity	linux64-shippable-qr	cold webrender	0.86 -> 0.94
8%	pinterest	Similarity2D	linux64-shippable-qr	cold webrender	0.92 -> 0.99

For up to date results, see: https://treeherder.mozilla.org/perfherder/alerts?id=27538

There's absolutely no way this test-only patch could have affected those tests. Can you please remove the reference to this bug from the alert?

Comment 13

[Alexandru Ionescu (needinfo me) [:alexandrui]]

(In reply to Kris Maglione [:kmag] from comment #12)

There's absolutely no way this test-only patch could have affected those tests. Can you please remove the reference to this bug from the alert?

The graph is pretty clear that this is the bug that caused the improvements. Could be indirectly. I did some retriggers to confirm. I will link the alert to another bug as soon as I find the cause.

Comment 14

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

The pushlog linked for the alert contains bug 1672597 (changes to image decoding) and the improvement is for Pinterest, an image heavy site. The improvement should be from that change (the change in this bug doesn't change the application, only a non-performance test).

Comment 15

[Alexandru Ionescu (needinfo me) [:alexandrui]]

Thanks, did some more retriggers.

Comment 16

[Marco Bonardo [:mak]]

I filed bug 1676506 for the Blob problem, thanks.

Comment 17

[Alexandru Ionescu (needinfo me) [:alexandrui]]

(In reply to Sebastian Hengst [:aryx] (needinfo on intermittent or backout) from comment #14)

The pushlog linked for the alert contains bug 1672597 (changes to image decoding) and the improvement is for Pinterest, an image heavy site. The improvement should be from that change (the change in this bug doesn't change the application, only a non-performance test).

The retriggers (see where's 9cbcab210 related to Alert 27538 - e301b90c2b0e) didn't reveal anything new. There's the possibility that this is just a spike and the cause to be a bit later, but not earlier. Did some more retriggers.

Comment 18

[Alexandru Ionescu (needinfo me) [:alexandrui]]

I'm sorry, :aryx, the retriggers didn't reveant anything new.

Comment 19

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

If one compares the first pinterest-vismet (old) and second one (newer), there is this difference:

old

[task 2020-10-27T19:48:08.914Z] INFO - Comparing videos to TASK_GROUP=OVoOn1h8Q7yuhB2B0QYAGg, TASK_ID=NeqmmcjiRHuaJDSsEGwqTQ
[task 2020-10-27T19:48:08.914Z] INFO - Downloading https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/NeqmmcjiRHuaJDSsEGwqTQ/artifacts/public/test_info/browsertime-results.tgz
[task 2020-10-27T19:48:09.698Z] INFO - Found 50 old videos
[task 2020-10-27T20:08:41.000Z] INFO - Average 3D similarity: 0.93241
[task 2020-10-27T20:08:41.000Z] INFO - Average 2D similarity: 0.99208

new

[task 2020-11-11T11:07:50.328Z] INFO - Downloading https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/P1dyBzJQSGunIMR5JvauPg/artifacts/public/test_info/browsertime-results.tgz
[task 2020-11-11T11:07:50.570Z] INFO - Found 50 old videos
[task 2020-11-11T11:29:53.070Z] INFO - Average 3D similarity: 0.8469
[task 2020-11-11T11:29:53.070Z] INFO - Average 2D similarity: 0.91059

Please check if the baseline (the different task ids) has a shift in performance metrics:

• old task
• new task

Comment 20

[Dave Hunt [:davehunt] [he/him] ⌚BST]

I've marked the improvement alert as invalid as the change was a simularity metric, which we shouldn't be alerting on. It also sounds like this wasn't the correct cause of the change. Alex: please don't clear needinfos without leaving a comment.