1261753 - "unexpected error" when logging in to accounts.firefox.com

Status: RESOLVED FIXED

(Cloud Services :: Server: Firefox Accounts, defect)

Description

[u549602]

Attachment: Screenshot_2016-04-04-09-53-52.png

Environment: Nightly
Device: Sony Xperia Z2 (Android 5.0.2 );
Build: Nightly 48.0a1 (2016-04-03);

Steps to reproduce:
1. Go to settings
2. Enter account credentials in order to log in
3. Try to sign in

Expected result:

User is signed in
Actual result:
User is not signed in and "Unexpected error" message is displayed

Notes: Please note that this issue occurs on all devices, phone and tablet

Comment 1

[Ryan Kelly [:rfkelly]]

Firefox Accounts is currently experiencing a partial service outage, the operations team is on the case but I don't yet have an expected time for return of service.

Latest service status is viewable at https://status.services.mozilla.com/

Comment 3

[:glob ✱]

it appears that the logging of this error to sentry is also failing, due to the code using GET instead of POST, resulting in "414 request too large".

Comment 4

[Justin Lazaro [:jlaz] (use needinfo)]

Bug 1261761 was reported in #MOC, and after an escalation to CloudOps, we've been notified that the team has been working on restoring service but with no definite ETA as of yet.

Comment 10

[Ryan Kelly [:rfkelly]]

We appear to be successfully back online, marking this RESOLVED/FIXED.

Comment 11

[Paul Silaghi, QA [:pauly]]

Now I got: "Attempt limit exceeded" error message.

Comment 12

[Paul Silaghi, QA [:pauly]]

Still doesn't work. Thoughts?

Comment 13

[Kevin Brosnan [Ex-Mozilla]]

[Tracking Requested - why for this release]: Release Management may be interested in this

Comment 14

[u549602]

On the mobile side, a work-around has been found to avoid the "Attempt limit exceeded" error message when trying to log in.

A hot-spot internet access has been created and if connected to it, i was able to log in to my sync account. 
When multiple devices are connected to a wi-fi and trying to log in with multiple sync accounts on different devices, the network generates only one exiting IP and maybe the server recognizes it as a conflict. 
P.S. Is there any way to hard delete the connection attempts from our IP?

Comment 15

[Ryan Kelly [:rfkelly]]

> Is there any way to hard delete the connection attempts from our IP?

There's no way for you to clear the rate-limiting yourself.  I can see what's possible on our end - please file an employee-confidential bug, list the IP addresses you need, and cc me.

Comment 16

[[SV Manager] Florin Mezei, QA (:FlorinMezei)]

Isn't this still an issue though? It seems to me we may be reaching this limit quite easily.

Comment 17

[Ryan Kelly [:rfkelly]]

Yes, we're also working on a better long-term solution to prevent this from happening so easily in the future, but unfortunately it's a tricky balancing act

Comment 18

[Ryan Kelly [:rfkelly]]

We've deployed some more fixes for the rate-limiting logic, and are seeing a much lower overall rate of "attempt limit exceeded" errors.  Can you confirm whether things are working better from your side?

Comment 19

[[SV Manager] Florin Mezei, QA (:FlorinMezei)]

Things seem to work fine now. I'll let you guys know if we run into additional problems. Thank you for working on this!

Comment 20

[Liz Henry (:lizzard) (relman/hg->git project)]

Sounds like this was fixed while 48 was still in nightly.

Comment 21

[[SV Manager] Florin Mezei, QA (:FlorinMezei)]

This was a server side fix... after ~1 month we've had no other issues reported with this.