Closed Bug 1897093 Opened 1 year ago Closed 18 days ago

Intermittent TEST-UNEXPECTED-TIMEOUT | comm/mail/base/test/browser/browser_mailContext.js | waiting for Gloda to finish indexing

Categories

(Thunderbird :: General, defect)

Product:
Thunderbird
Component:
General
Type:
defect

Tracking

(thunderbird143 fixed)

Status:
RESOLVED FIXED
Milestone:
143 Branch
Tracking Flags:
Tracking Status
thunderbird143 --- fixed

People

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

Details

(Keywords: intermittent-failure, intermittent-testcase, test-disabled)

Attachments

(4 files)

crash backtrace from treeherder log
30.67 KB, text/plain
Details
Bug 1897093 - try to fix intermittent comm/mail/base/test/browser/browser_mailContext.js failure. r=#thunderbird-reviewers
48 bytes, text/x-phabricator-request
Details | Review
Bug 1897093 - Fix contextmenu event when the current row is out of view. r=#thunderbird-reviewers
48 bytes, text/x-phabricator-request
Details | Review
Bug 1897093 - Improve performance of Gloda indexing in tests. r=#thunderbird-reviewers
48 bytes, text/x-phabricator-request
Details | Review

Oh, reading up a bit higher in the log, it looks like there was just some sort of test-timeout, and so we got force-killed to interrupt that. So the crash was an intentional crash, from the harness killing us.

Here's the paper trail (notice "Killing process: 1051" followed by "PROCESS-CRASH" for PID 1051):

[task 2024-05-16T00:15:35.335Z] 00:15:35     INFO - TEST-UNEXPECTED-TIMEOUT | comm/mail/base/test/browser/browser_mailContext.js | application timed out after 370 seconds with no output
[task 2024-05-16T00:15:35.335Z] 00:15:35     INFO - TEST-INFO took 390830ms
[task 2024-05-16T00:15:35.335Z] 00:15:35     INFO - Buffered messages finished
[task 2024-05-16T00:15:35.336Z] 00:15:35  WARNING - Force-terminating active process(es).
[...]
[task 2024-05-16T00:15:35.465Z] 00:15:35     INFO - Killing process: 1051
[...]
[task 2024-05-16T00:15:46.072Z] 00:15:46     INFO - PROCESS-CRASH | application crashed [@ mozilla::nsDisplayListCollection::~nsDisplayListCollection] | comm/mail/base/test/browser/browser.ini 
[task 2024-05-16T00:15:46.072Z] 00:15:46     INFO - Process type: main
[task 2024-05-16T00:15:46.072Z] 00:15:46     INFO - Process pid: 1051

So this was really just a test-timeout, not a crash.

One thing I notice in the screenshot is a popup saying "Your disk is almost full":
https://firefoxci.taskcluster-artifacts.net/PDydVPTmQRamwSgzSlQajg/0/public/test_info/mozilla-test-fail-screenshot_45izsj11.png

Not sure how close we were to being full (and I do see "disk_size": "233.47 GiB", at the top of the log which seems large). But I wonder if that might be associated with what made us timeout?

Summary: Intermittent application crashed [@ mozilla::nsDisplayListCollection::~nsDisplayListCollection] | single tracking bug → Intermittent TEST-UNEXPECTED-TIMEOUT | comm/mail/base/test/browser/browser_mailContext.js | single tracking bug

--> Bouncing over to Thunderbird|General, ni=mkmelin for any further classification he wants to do here.

Component: Layout → General
Flags: needinfo?(mkmelin+mozilla)
Product: Core → Thunderbird

Thanks!

Flags: needinfo?(mkmelin+mozilla)

I'm going to disable this test on macOS debug, as it fails there far too often. For some reason the whole program locks up if we open too many context menus in quick succession.

Keywords: crash → leave-open, test-disabled
Pushed by geoff@darktrojan.net: https://hg.mozilla.org/comm-central/rev/2292ac38a58f Disable browser_mailContext.js on macOS debug. rs=me

Pushed by mkmelin@iki.fi:
https://hg.mozilla.org/comm-central/rev/05953ce3ebe2
try to fix intermittent comm/mail/base/test/browser/browser_mailContext.js failure. r=tobyp

Keywords: checkin-needed-tb

The current spate of failures is happening because we get here without a row and with retry true. At that point it just gives up.

I don't know why it's happening, the scrollToIndex call should result in a row being ready before the end of the event loop, i.e. before the requestAnimationFrame callback runs. I've tried replacing it with setTimeout and that fails too sometimes.

Keywords: checkin-needed-tb

Pushed by mkmelin@iki.fi:
https://hg.mozilla.org/comm-central/rev/45efc0f59964
Fix contextmenu event when the current row is out of view. r=mkmelin

Keywords: checkin-needed-tb

"waiting for Gloda to finish indexing" seems to be the only failure mode happening regularly, so I'm changing the bug title to that to avoid missing new failures.

Summary: Intermittent TEST-UNEXPECTED-TIMEOUT | comm/mail/base/test/browser/browser_mailContext.js | single tracking bug → Intermittent TEST-UNEXPECTED-TIMEOUT | comm/mail/base/test/browser/browser_mailContext.js | waiting for Gloda to finish indexing

I know what's causing Gloda indexing to be so incredibly slow on macOS 14. It's called "macOS 14".

Assignee: nobody → geoff
Status: NEW → ASSIGNED
Crash Signature: [@ mozilla::nsDisplayListCollection::~nsDisplayListCollection]

This is a bizarre bug. The tests I've changed here (mostly the first one) are failing when run
on macOS 14.70. I've tracked it down to this part of the MIME parser where it goes to the OS and
asks for a file extension for a given content type. The OS is responding very slowly and since
it's getting asked the same thing many times, it's taking too long and the test is timing out.

I've hard-coded in the extension for "text/plain" is "txt". Because it is.

Let's open a new bug if we get any more intermittent failures from this test. We've already made three attempts at fixing various things.

Keywords: leave-open → checkin-needed-tb
Target Milestone: --- → 143 Branch

Pushed by corey@thunderbird.net:
https://hg.mozilla.org/comm-central/rev/0ef2997da726
Improve performance of Gloda indexing in tests. r=aleca

Status: ASSIGNED → RESOLVED
Closed: 18 days ago
Keywords: checkin-needed-tb
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: