Closed Bug 1258897 Opened 6 years ago Closed 2 years ago

Intermittent test_ext_sendmessage_reply2.html | Test timed out. | Extension left running at test shutdown


(WebExtensions :: General, defect, P2)



(firefox56 disabled, firefox57 disabled)

Tracking Status
firefox56 --- disabled
firefox57 --- disabled


(Reporter: KWierso, Assigned: rpl)


(Blocks 1 open bug)


(Keywords: intermittent-failure, Whiteboard: triaged [stockwell fixed])


(2 files)

Whiteboard: triaged
Closed: 6 years ago
Resolution: --- → DUPLICATE
Duplicate of bug: 1268198
Resolution: DUPLICATE → ---
Depends on: 1274775
Bulk assigning P3 to all open intermittent bugs without a priority set in Firefox components per bug 1298978.
Priority: -- → P3
Component: WebExtensions: Untriaged → WebExtensions: General
This intermittent has not occurred for over 180 days, so closing.
Closed: 6 years ago5 years ago
Resolution: --- → WORKSFORME
This long-dormant failure re-started around July 28 and has been reported frequently since then. It seems like all platforms are affected, but Windows 7 is worst.

:andym - Can you find someone to investigate this frequent intermittent test failure?
Flags: needinfo?(amckay)
Whiteboard: triaged → triaged [stockwell needswork]
Pushed by
disable test_ext_sendmessage_reply2.html. r=me a=testonly
leaving this open, please consider looking into the failure and fixing this in the near future :)
Keywords: leave-open
Whiteboard: triaged [stockwell needswork] → triaged [stockwell disabled]
This is skipped on all platforms, so should be treated as a high priority to re-enable. Marking as P2.
Flags: needinfo?(amckay)
Priority: P3 → P2
Product: Toolkit → WebExtensions

This patch should fix the test_ext_sendmessage_reply2.html intermittency which, based on the
timeout failure that I've been able to intermittently reproduce when running this test
locally using --verify, seems to be likely due to the following race:

the tabs.create API (used to create the extension tabs used as part of the tests) was called
right before subscribing the extension message listeners, and so from time to time the
extension tabs were not able to successfully send their messages to the background page listeners
(because the background page didn't subscribe the listeners yet) and as a result the test was getting
stuck waiting for extension messages that were never going to be received.

Assignee: nobody → lgreco
Pushed by
Fix test_ext_sendmessage_reply2.html intermittency and re-enable. r=mixedpuppy

This test seems to be still failing intermittently, and its intermittency rate looks high enough.

Looking to the logs collected on the recent intermittent failures, it seems that the issue
is still related to the extension tabs opened in the test:

The two extension tabs (each one part of one of the two test extension) are going to exchange messages
between each other, and currently there is a chance (and apparently a very good chance) that one of the
two extension tabs is not yet ready to listen to the messages sent from the other extension tab.

This patch applies to the test the additional changes needed to ensure that both the extension tabs
are ready to exchange messages, before we let them exchange these messages over the extension messaging

Pushed by
Ensure that both test_ext_sendmessage_reply2 tab scripts are ready to exchange messages. r=mixedpuppy

There are 19 total failures in the last 7 days.

Luca can you please take another look at this?
Most recent failure log:

Flags: needinfo?(lgreco)
Whiteboard: triaged [stockwell disabled] → triaged [stockwell needswork:owner]

(In reply to Andreea Pavel [:apavel] from comment #30)

There are 19 total failures in the last 7 days.

Luca can you please take another look at this?
Most recent failure log:

Hi Andreea,
I was already keeping an eye on this test after we landed the first patch that has re-enabled it (linked in comment 24) and, as I mentioned in comment 25, I noticed that the test was still failing intermittently and its rate was clearly increasing and so I took care of it right away.

The patch attached in comment 25 has been already landed (comment 27) and looking at the failure we got since then:

the last patch seems to have been able to prevent further failures from being triggered (and the last hg revisions with a failure, the one linked in comment 30 and the last failure happened on try on Feb 2, do not include the changes from the last patch).

Flags: needinfo?(lgreco)
Whiteboard: triaged [stockwell needswork:owner] → triaged [stockwell fixed]

Luca, thanks for clarifying. I saw you landed the patch, however as Bogdan mentioned, there still was a failure on the push with your patch, thus I wanted to clarify this with you.

Looking in orange factor now, i see that there are no recent failures so indeed this seems to be fixed.

Closing as fixed (as mentioned in Comment 32 and Comment 33 there are no further failure from this intermittent starting from Feb 2 2019)

Closed: 5 years ago2 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.