902177 - When committing an edit, got an empty page

Status: RESOLVED WORKSFORME

(developer.mozilla.org Graveyard :: General, defect)

Description

[Jean-Yves Perrier [:teoli]]

What did you do?
================
1. Edited a page https://developer.mozilla.org/en-US/docs/Web/API/NodeIterator.root
2. Clicked the "Save" button

What happened?
==============
A blank page with the URL https://developer.mozilla.org/en-US/docs/Web/API/NodeIterator.root$edit is displayed

What should have happened?
==========================
A newly generated page with https://developer.mozilla.org/en-US/docs/Web/API/NodeIterator.root

Is there anything else we should know?
======================================
I have logged the request/response I got:

Request URL: 	https://developer.mozilla.org/en-US/docs/Web/API/NodeIterator.root$edit
Request Method: 	POST

Status Code: 	HTTP/1.1 302 FOUND

Request Headers 22:28:56.000
User-Agent:	Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:25.0) Gecko/20100101 Firefox/25.0
Referer:	https://developer.mozilla.org/en-US/docs/Web/API/NodeIterator.root$edit
Host:	developer.mozilla.org
Connection:	keep-alive
Accept-Language:	en-US,en;q=0.5
Accept-Encoding:	gzip, deflate
Accept:	text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8

Sent Cookie
WT_FPC:	id=207.218.72.66-87408320.30250580:lv=1365589779322:ss=1365589779322
sessionid:	b05479f898fb6697226c0f26130f4d69
optimizelySegments:	{"237061344":"ga-201%2Bcourse","237321400":"ff","237335298":"search","237485170":"false"}
optimizelyEndUserId:	oeu1369378799859r0.14241542622644765
optimizelyCustomEvents:	{"oeu1369378799859r0.14241542622644765":["edit_page_button"]}
optimizelyBuckets:	{"239795743":"239955500"}
dloadday:	207.218.72.65.1348403373339811
csrftoken:	bb236cf4b4172e95ad7c1e50e9726005
browserid_explained:	1
__utmz:	194606776.1375733670.484.84.utmcsr=google|utmccn=(organic)|utmcmd=organic|utmctr=(not provided)
__utmz:	150903082.1373452366.3.2.utmcsr=google|utmccn=(organic)|utmcmd=organic|utmctr=(not provided)
__utmxx:	150903082.uPnMIIvQRCyx9V6nTIQ-Bw$71153379-20:1368773310:15552000
__utmx:	150903082.uPnMIIvQRCyx9V6nTIQ-Bw$71153379-20:0
__utmc:	194606776
__utmb:	194606776.7.10.1375823936
__utma:	194606776.1999771830.1362062484.1375773252.1375823936.487
__utma:	150903082.1084948875.1368773311.1369556158.1373452366.3

Response Headers Δ1311ms
X-Robots-Tag:	noindex
X-Frame-Options:	DENY
X-Cache-Info:	not cacheable; request wasn't a GET or HEAD
X-Backend-Server:	developer2.webapp.scl3.mozilla.com
Vary:	Cookie
Server:	Apache
Location:	https://developer.mozilla.org/en-US/docs/Web/API/NodeIterator.root
Keep-Alive:	timeout=5, max=1000
Date:	Tue, 06 Aug 2013 21:28:56 GMT
Content-Type:	text/html; charset=utf-8
Content-Length:	0
Connection:	Keep-Alive
Received Cookie
multidb_pin_writes:	y


This problem is happening since last week-end, not before Friday.

Comment 1

[Jean-Yves Perrier [:teoli]]

This is happening intermittently, I would say 40% of time (need to gather more statistic).

Note that the changes of the page always are committed successfully.

Feel free to ask any other information (do you think to check if X-Backend-Server is always the same is useful?) I am eager to help you solve this problem.

Comment 2

[Jean-Yves Perrier [:teoli]]

Tonight for the moment, I made 6 commits (=6 edits) and I got 100% of the time this symptom.

This is more than usual.

I will continue to check.

Comment 3

[Jean-Yves Perrier [:teoli]]

Looked at several requests and also got these backend servers:
X-Backend-Server:	developer1.webapp.scl3.mozilla.com
X-Backend-Server:	developer2.webapp.scl3.mozilla.com
X-Backend-Server:	developer3.webapp.scl3.mozilla.com

Note that tonight, this specific symptom happened about 25 times, that is 100% of then time (when clicking the save button).

It wasn't my experience this morning, where it happened often but not always.

Comment 4

[Les Orchard [:lorchard]]

This doesn't involve KumaScript. It probably involves the Celery queue issues we've been having. Changing the blocker.

I edited the page 5 times in a row without seeing the issue. Thinking possibly changes we made yesterday improved things? 

Reopen this bug, if it happens again. If it does, and we have an overflowing Celery work queue, then we definitely know that's the problem.

Comment 5

[Les Orchard [:lorchard]]

FWIW, I successfully edited the page in question several times today:

https://developer.mozilla.org/en-US/docs/Web/API/NodeIterator.root$history

Comment 6

[Jean-Yves Perrier [:teoli]]

This is definitively still happening. The three first edits this morning lead to blank page (with the edit committed):

https://developer.mozilla.org/en-US/docs/Web/HTML/Element/blink Aug 7, 2013 10:14:41 PM
https://developer.mozilla.org/en-US/docs/Web/HTML/Element/blink Aug 7, 2013 10:14:19 PM
https://developer.mozilla.org/en-US/docs/Web/HTML/Element/blink Aug 7, 2013 10:13:37 PM

(Kuma times)

I will keep a list.

Comment 7

[Jean-Yves Perrier [:teoli]]

https://developer.mozilla.org/en-US/docs/Web/HTML/Element/blink Aug 7, 2013 10:18:43 PM
https://developer.mozilla.org/en-US/docs/Web/HTML/Element/blink Aug 7, 2013 10:22:31 PM
https://developer.mozilla.org/en-US/docs/Mozilla/Firefox/Releases/23 Aug 7, 2013 10:23:42 PM

Then, from then (that is a few minutes later, but before 10:45pm), the problem vanished. If I get it again today, I will report.

(I keep this open as it still need investigation as I got it after it was officially fixed and we have no explanation for that).

Comment 8

[John Karahalis [:openjck]]

Les has reason to believe this is fixed permanently. Please reopen if it happens again.

Comment 9

[Jean-Yves Perrier [:teoli]]

Please close it as WORKSFORME, not as INVALID, it was a perfectly valid bug.

Comment 10

[John Karahalis [:openjck]]

My mistake. I meant to hit WORKSFORME.

Comment 11

[John Karahalis [:openjck]]

Looks like this is happening again.

See https://groups.google.com/forum/#!topic/mozilla.mdn.drivers/Qd7dubaGXUg

Comment 12

[John Karahalis [:openjck]]

Jean-Yves: Luke is wondering -- did this happen right after hitting "Save" or did you have to wait for a while before this empty page came up? Is there anything else you can tell us about this recent experience?

Comment 13

[Jean-Yves Perrier [:teoli]]

This happen right after the save. I wouldn't say instantaneously, but definitively much quicker than usual.

The commit is successfully done, the URL in the address bar stay ...$edit (no redirect to the final page)

Note that since the last occurrence linked by John, it didn't happened again.

Comment 14

[Luke Crouch [:groovecoder]]

I had WebOps check the celery queue to make sure we weren't getting into another overloaded situation which caused this. The celery queue looks fine and nothing on the backend has changed recently.

Raymond, can you do some exploratory edits on MDN to try to reproduce this and discover the STR?

Comment 15

[raymond [:retornam] (inactive)]

(In reply to Luke Crouch [:groovecoder] from comment #14)
> I had WebOps check the celery queue to make sure we weren't getting into
> another overloaded situation which caused this. The celery queue looks fine
> and nothing on the backend has changed recently.
> 
> Raymond, can you do some exploratory edits on MDN to try to reproduce this
> and discover the STR?

Do you want me to test on dev or stage?

Comment 16

[Luke Crouch [:groovecoder]]

Please test on stage and prod in a custom wiki namepsace (e.g., under User:retornam) - it may be related to celery which has a different setup between dev and stage & prod.

Comment 17

[Jean-Yves Perrier [:teoli]]

Adding here the comment I posted on mdn-driver last Saturday:
> Ok, something quite weird is happening with this one.

> It happened 100% of the time again, until I restarted my Nightly.

> So if somebody else get a blank page, try to see if :
> 1) It happens 100% of the time when commiting an edit.
> 2) It goes away when you restart your browser.

> Also I'm interested to know if other people experience this one from
> time to time or if I have to investigate on my side (though I don't have
> a lot of extensions)

> Have a nice Sunday,

Comment 18

[Les Orchard [:lorchard]]

Doesn't seem reproducible right now. At what URL did this happen last? We might be able to search error logs for Saturday, given the URL.

Comment 19

[Jean-Yves Perrier [:teoli]]

Unfortunately I didn't wrote it down.

I propose we put this bug on hold until it happens again. Since the last occurrence of Saturday, with 100s of editions done this week, I didn't got it once. (Nor any other stability problem).

If it happens again, I will reopen the bug (and note the precise time and page of the first occurrence ;-) )

Thanks for the help.

Comment 20

[raymond [:retornam] (inactive)]

I haven't been able to reproduce this. I made several empty edits.

https://developer.allizom.org/en-US/docs/tools/food

Comment 21

[Janet Swisher]

Possibly related to bug 866536?

Comment 22

[Jean-Yves Perrier [:teoli]]

I think it is a different problem. The URL in bug 866536 is clearly wrong (leading to a site problem).

In this bug, the REDIRECT didn't happen but all the rest looked correctly.

Also as I saw it once on a different site, I think it is (was?) a local problem.