Wikis were dead for several minutes on 2024-12-05
Closed, ResolvedPublicBUG REPORT
Actions

Assigned To

None

Authored By

	Wellverywell
	Dec 5 2024, 4:40 AM

Description

Wikis are broken since at least 04:30 UTC (probably even earlier, according to Wikimedia Status it started around 04:15-04:20).

Steps to replicate the issue (include links if applicable):
Try to open any Wikipedia or Commons.

What happens?:
Either this:

Request from myip.myip.myip.myip via cp5021.eqsin.wmnet, ATS/9.2.6
Error: 502, Broken pipe at 2024-12-05 04:19:21 GMT

Or this:

Original error: upstream connect error or disconnect/reset before headers. reset reason: connection failure, transport failure reason: delayed connect error: 111

What should have happened instead?:
Wikis should open normally

Event Timeline

Wellverywell created this task.Dec 5 2024, 4:40 AM

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptDec 5 2024, 4:40 AM

Wellverywell updated the task description. (Show Details)Dec 5 2024, 4:41 AM

SCP-2000 subscribed.Dec 5 2024, 4:43 AM

Currently they either do not open at all (as described above), or open very slowly.

Screenshot 2024-12-05 at 10.16.42 AM.png (996×1 px, 95 KB)

getting this on every wiki.

Original error: upstream connect error or disconnect/reset before headers. reset reason: connection failure

I observed that issue is on all subdomains like ta.wikipedia.org or en.wikisource.org, etc.,

First error message I received:
Request from [nnn.nnn.nnn.nnn] via cp4037.ulsfo.wmnet, ATS/9.2.6
Error: 502, Broken pipe at 2024-12-05 04:13:26 GMT

Wellverywell updated the task description. (Show Details)Dec 5 2024, 4:55 AM

The error rate seems to be decreasing. https://downdetector.com/status/wikipedia/

Discussion @ https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(technical)#Login_errors

I even got an error when trying to log in to Phabricator!

Seems it became better.

Access is ok at the moment. The problem had occurred for at least 40 minutes. I couldn't report the problem on Phab at the time, because I couldn't get to unified login!

@Nurg -- just a heads-up, you revealed your IP in the previous comment.

https://www.wikimediastatus.net/incidents/3g2ckc7bp6l9

It may be related to my mass upload task to Commons: https://commons.wikimedia.org/wiki/Special:Contributions/MBHbot

If it starts again, you could log in "mbh" Toolforge account and kill one-off task named "upload" or "scripts". I am not at home now and can't log into Toolforge console from phone. I have asked other owners of my tool to kill this process. It should upload ~20 000 files after this point.

UPD: Task has stopped.

Pppery unsubscribed.Dec 5 2024, 5:36 AM

Aklapper renamed this task from Wikis are dead for several minutes to Wikis were dead for several minutes on 2024-12-05.Dec 5 2024, 11:43 AM

Aklapper added a project: Wikimedia-Incident.

	F57780640: Screenshot 2024-12-05 at 10.16.42 AM.png
	Dec 5 2024, 4:49 AM

Wikis were dead for several minutes on 2024-12-05Closed, ResolvedPublicBUG REPORTActions

Description

Event Timeline

Wikis were dead for several minutes on 2024-12-05
Closed, ResolvedPublicBUG REPORT
Actions