Closed Bug 1673534 Opened 4 years ago Closed 3 years ago

twitter.com won't load NS_BINDING_ABORTED

Categories

(Core :: DOM: Service Workers, defect, P2)

Firefox 84
Desktop
Windows 10
defect

Tracking

()

RESOLVED WORKSFORME
Tracking Status
firefox-esr78 --- unaffected
firefox82 --- wontfix
firefox83 --- wontfix
firefox84 --- wontfix
firefox85 --- wontfix
firefox86 --- wontfix
firefox87 --- wontfix
firefox88 --- wontfix
firefox89 --- wontfix
firefox90 --- fix-optional
firefox91 --- fix-optional

People

(Reporter: alice0775, Assigned: asuth)

References

(Regression)

Details

(Keywords: nightly-community, regression)

Attachments

(3 files)

Attached image screenshot

Reproducible: sometimes

Steps to reproduce

  1. Open https://twitter.com/

Actual Results:
Contents area is blank. The page loading indicator continues to be displayed.
Network monitor indicates NS_BINDING_ABORTED

Attached file about:support

I couldn't manage to reproduce it after many tries on Windows 10 x64, MacOS 10.14 and on Ubuntu 20.04.
Assigning "Core: Networking: Cache" component it could be the right one if not please change it.

Component: Untriaged → Networking: Cache
Product: Firefox → Core

Could you try to make a http log?
Can you reproduce this with a clean profile?

Thanks.

Flags: needinfo?(alice0775)
Attached file log.zip

(In reply to Kershaw Chang [:kershaw] from comment #3)

Could you try to make a http log?
Can you reproduce this with a clean profile?

Thanks.

I attached the Log using about:networking.

Flags: needinfo?(alice0775)

This looks like bug 1658192.
Could you try to unregister the service worker from twitter.com?
https://support.mozilla.org/en-US/kb/twitter-isnt-working-firefox

Flags: needinfo?(alice0775)

(In reply to Kershaw Chang [:kershaw] from comment #5)

This looks like bug 1658192.
Could you try to unregister the service worker from twitter.com?
https://support.mozilla.org/en-US/kb/twitter-isnt-working-firefox

Yes, the workaround fixes this issue temporary on Noghtly84.0a1.

Flags: needinfo?(alice0775)
Status: NEW → RESOLVED
Closed: 4 years ago
Resolution: --- → DUPLICATE

I think that same root cause. But regression window is different.

Steps:

  1. Start browser with new profile
  2. Open https://twitter.com/ and bookmark this
  3. about:serviceworkers and unresister
  4. Restart Browser
  5. Middle click the bookmark in new tab (optionally open it again )
  6. Restart Browser
  7. Middle click the bookmark (optionally open it again )

Actual results:
No longer load the page

Regression window:
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=d01b9d7154d2557ab817e58cb0f143ba4eb34653&tochange=49008be9966973f89e6608355003ac280580c44c

Before landing Bug 1662925 , 1st attempt to open https://twitter.com/, Network is blocked, 2nd attempt succsess
After landing Bug 1662925 , Once the block has occurred, it will continue to be blocked until SW is unresisted. And after closing the browser, Bug 1675068 occurs.

Status: RESOLVED → REOPENED
Has Regression Range: --- → yes
Component: Networking: Cache → DOM: Service Workers
Regressed by: 1662925
Resolution: DUPLICATE → ---
Status: REOPENED → NEW
Has STR: --- → yes
Assignee: nobody → bugmail
Severity: -- → S3
Status: NEW → ASSIGNED
Priority: -- → P2

Andrew, what's the status on this regression?

Flags: needinfo?(bugmail)

I'm running Firefox 87 beta 6 and this bug affects pretty much all websites.

I have to restart the web browser each ~two hours because it stops opening any websites altogether.

I get multiple NS_BINDING_ABORTED errors in the network log.

I'm using the official Firefox release from Mozilla's FTP on Fedora 33 with all updates applied.

Is there any quick workaround?

Firefox 87 beta 7 is still affected.

(In reply to Artem S. Tashkinov from comment #10)

I have to restart the web browser each ~two hours because it stops opening any websites altogether.

This sounds like it might be a different problem if it's impacting all sites, not just ones that have ServiceWorkers registered in about:serviceworkers. In particular, there might be a problem involving web extensions that use webRequest to intercept all network requests? It's probably worth running in safe mode to help rule that out.

I get multiple NS_BINDING_ABORTED errors in the network log.

It's worth double checking these against about:serviceworkers too.

(In reply to Artem S. Tashkinov from comment #11)

Is there any quick workaround?

Noting that you may be experiencing a different situation above and what I'm describing is only relevant to this bug:

There are 4 classes of mitigations:

  1. To bypass ServiceWorkers for a single page (re)load.
    • Hold down shift when reloading a page to force bypassing ServiceWorkers. This means holding shift while clicking the reload/refresh icon, or adding shift to the key combination you use to reload. For example, shift-control-R on linux.
  2. Unregister a ServiceWorker for a site so that it can reinstall a fresh, up-to-date copy, but without clearing any data the ServiceWorker has stored locally as offline data in Firefox. Note that a site could potentially also be experiencing problems related to the disk quota it's allowed to use. In that case you potentially want to see the next option.
    • If you're experiencing issues with a single site or a small number of sites, you can open about:serviceworkers in a tab and find the site in question and click the "unregister" button to remove its ServiceWorker. The next navigation won't involve a ServiceWorker, and the site will likely reinstall the ServiceWorker.
  3. Clear the data for the site/origin. This will uninstall all ServiceWorkers for the site and all of their related data. This helps deal with the situation where the site has used up all the storage it is allowed. This won't impact any data the site maintains on its servers, but will impact any offline data it has cached locally.
  4. Disable installation of ServiceWorkers and remove all existing ServiceWorkers but without clearing offline data. Note that this will disable push notifications from sites.
    • Set the preference dom.serviceWorkers.enabled to false in about:config. This will prevent new ServiceWorkers from being registered.
    • Remove existing ServiceWorker registrations. 2 options here that don't involve clearing all site data. (If you clear all offline data, all ServiceWorkers will be cleared as part of the process.)
      1. Delete the file serviceworker.txt from your profile directory when Firefox is not running. You can find your profile directory via opening about:profiles in a tab. (There's no harm in deleting the file when Firefox is running, it just might not accomplish the goal of clearing the list because Firefox may write a new copy of the data to disk on shutdown.)
      2. Bring up about:serviceworkers and click unregister on every registration.

(In reply to Andrew Sutherland [:asuth] (he/him) from comment #12)

Great many thanks for your comment.

I've noticed the combination of these three add-ons wrecks havoc and causes multiple websites to fail with this error.

They are:

https://addons.mozilla.org/firefox/addon/canvasblocker
https://addons.mozilla.org/firefox/addon/clean-links-webext
https://addons.mozilla.org/firefox/addon/ublock-origin

After disabling all three and opening affected websites in a new tab, Firefox starts working again at which point I can reenable them. That's all kinda weird.

Firefox 88.0b8

Self-contained reproducing code

  1. Go here https://code.quarkus.io/?e=resteasy-qute
  2. You must have Java 8 or newer
  3. Extract archive somewhere
  4. Open terminal and run this ./mvnw compile quarkus:dev (for linux, for windows use mvnw.cmd)
  5. Wait a bit until it starts; it must print something like this: Installed features: [cdi, qute, resteasy, resteasy-qute]
  6. Open http://localhost:8080/qute/quarks
  7. Click on Generate a new quark, new circle should appear
    Works fine on Firefox 87, does not work on Firefox 88.0b8, FF 87 was tested with existing profile, dev. edition was tested with new clean profile

Generate new quart link code:
<a href="/qute/quarks" onclick="fetch('/qute/quarks/add', { method: 'POST' });">Generate a new quark</a>

In network info i see NS_BINDING_ABORTED error

Unregistering of all serviceworkers (go to about:serviceworkers - unregister each item) does NOT work for me.

reproducability: near 100% on my local machine after PC reboot (but yesterday it unhangs after a few hours, now it hangs again)
after auto-update to 90.0.1
run firefox and open any site (not help: reloading firefox, reloading PC, disabling avast antivirus, chaning )

I have posted on https://bugzilla.mozilla.org/show_bug.cgi?id=1658192 too

Hi Alice, are you still seeing this?

Flags: needinfo?(bugmail) → needinfo?(alice0775)

(In reply to Jens Stutte [:jstutte] from comment #16)

Hi Alice, are you still seeing this?

I cannot reproduce this anymore in Firefx91esr, 97.0.1 as well as Nightly99.0a1.

Flags: needinfo?(alice0775)

Great, thanks for verifying! If anybody else still encounters this, please do not hesitate to report it back. Thanks for your support.

Status: ASSIGNED → RESOLVED
Closed: 4 years ago3 years ago
Resolution: --- → WORKSFORME

I maybe intermittently experience this, but with Gmail. I'm currently on the latest nightly, 20220226213636, on a Windows 7 system but I've had it happen previously at least a few times over the last month.

I have a pinned tab for Gmail open all the time. What I experience is Gmail saying it's not connected and it never recovers. If I refresh the page, all the requests fail with NS_BINDING_ABORTED except a few blocked by uBlock Origin.

If I update nightly while Firefox is in this state, then after the restart, Gmail works again. After searching BZ for NS_BINDING_ABORTED, I found this bug, so I tried unregistering mail.google.com's service workers and refreshing the tab, and then Gmail worked. I haven't tried disabling uBlock Origin as the user in comment #13 mentioned. I think I've tried ctrl-shift-r to refresh the tab, but I don't think that alone worked. I'll try that again if the problem reoccurs.

Please let me know what additional info would be helpful or if I should file a separate bug. I could possibly not update nightly every day to see if it'd make the problem occur for me more often.

You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: