Goffi Reporter
	Description • 8 years ago

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:43.0) Gecko/20100101 Firefox/43.0 Iceweasel/43.0.4
Build ID: 20160107230306

Steps to reproduce:

Hello,

When using Firefox with our application, we can see in devtools/network tab that many long polling XMLHttpRequest (post request) are aborted without apparent reason (reproducible even on localhost).

The requests works at the beginning, the server is reached, then after a short period of time (usually less than one minute), they are aborted (status code of 0). There is no timeout on the requests, and not abort() call in the code.

This issue happen with Firefox 43 (and also tried with Firefox/Iceweasel 44.0~b6-1 from Debian experimental) even with --safe-mode. The issue doesn't happen with Firefox 42, Firefox 38, and Chromium 48.0.2564.23.

Our website can be tested at https://www.libervia.org (I have created the account mozilla/mozilla so you can test).

I join a screenshot of the network tab in devtools, where json_signal_api request are aborted.

Tested on Debian/Sid, Firefox/Iceweasel 43.0.4-1 (and 44.0~b6-1 from experimental).

Thanks


Actual results:

Request are aborted randomly (but really often) after a random but short (less than 1 min) amount of time.


Expected results:

request should be pending until an answer is received

	Loic
	Comment 1 • 8 years ago

It could be a dupe of bug 1234548 (even if I'm not expert), which is going to be fixed.

Could you test Nightly in a few days when the patch landed, please.
https://nightly.mozilla.org/

	Goffi Reporter
	Comment 2 • 8 years ago

Thanks for you reply. The issue seems indeed similar, but disabling IPv6 as suggested in comments doesn't change anything for me.

I'll try in a couple of day. How can I know when the patch has landed into nightly build?

Thanks

	Loic
	Comment 3 • 8 years ago

The patch should be in the Nightly of the day, just wait for the end of the day (near 17-18h00 UTC) to download and test it.

	Goffi Reporter
	Comment 4 • 8 years ago

hello,

I've just tried with the last nightly build, it's working now the bug is fixed.

thanks !