From the block log:

08:59, 14 October 2015 Multichill (talk | contribs) blocked Selenium user (talk | contribs) with an expiry time of indefinite (autoblock disabled) (Unauthorized bot uploading junk)

Maarten, we would need that block to be lifted :-D

Per IRC discussion on Commons I sent a notice to the admin board: https://commons.wikimedia.org/w/index.php?title=Commons:Administrators%27_noticeboard&diff=175658099&oldid=175635898

@Steinsplitter kindly unblocked the bot. From the admin board:

I just unblocked the user per https://phabricator.wikimedia.org/T115465 . Can you please use {{tl|Test upload}} on the file (white in the information template) and we should probably rename the account to [[User:Selenium user (WMF QA bot)]] per suggestion on AN.
--[[User:Steinsplitter|Steinsplitter]] ([[User talk:Steinsplitter|<span class="signature-talk">{{int:Talkpagelinktext}}</span>]]) 13:24, 14 October 2015 (UTC)

Ah, the mysterious operator is Hashar! Another case where the peep system worked. I'll follow up in the other bug.

@Multichill Yup the job has been created a while back to ensure UploadWizard is working properly. We had some awkward issues that caused much havoc for commons users. The job pretty much guarantee UploadWizard works as attended and is taken care of by Release-Engineering-Team + Multimedia (if that later team still exists).

From QA list https://lists.wikimedia.org/pipermail/qa/2015-October/002411.html

Zeljko confirmed on the Commons village pump that we were no more using the 'Selenium User': https://commons.wikimedia.org/w/index.php?title=Commons%3AVillage_pump&type=revision&diff=175068085&oldid=175058874

That stands true for browser tests, but we missed the UploadWizard suite which is a different kind of job.

So blocking the user was definitely the correct course of action and that has let us identify the source of the junk files.

Why isn't it running on the Beta Cluster?

Why isn't it running on the Beta Cluster?

There is one on the beta cluster: https://integration.wikimedia.org/ci/job/UploadWizard-api-commons.wikimedia.beta.wmflabs.org/

But we need one to run against production to make sure it is definitely working on prod as well. It is more like a monitoring system.

In T115465#1726287, @hashar wrote:

Why isn't it running on the Beta Cluster?

There is one on the beta cluster: https://integration.wikimedia.org/ci/job/UploadWizard-api-commons.wikimedia.beta.wmflabs.org/

But we need one to run against production to make sure it is definitely working on prod as well. It is more like a monitoring system.

I disagree. If we can't use Beta Cluster / Staging to tell us whether the code works, we need to improve staging. :-) Making changes to the production site to prove it works isn't something we allow for e.g. edits, and shouldn't be allowed for uploads.

In T115465#1726331, @Jdforrester-WMF wrote:

Making changes to the production site to prove it works isn't something we allow for e.g. edits, and shouldn't be allowed for uploads.

Yes we do... https://en.wikipedia.org/wiki/Special:Contributions/Ops_Monitor_%28WMF%29

I disagree. If we can't use Beta Cluster / Staging to tell us whether the code works, we need to improve staging. :-) Making changes to the production site to prove it works isn't something we allow for e.g. edits, and shouldn't be allowed for uploads.

It is not only about validating the code and API entry point, but also validating that the stashing and object storage backend (Swift) honors the requests.

I have reblocked the account. The account is still editing for some days now. The account can be unblocked when the bugs are fixed.

Surely blocking the user again is a constructive way to prioritize the work that needs to be done on T115469 ? I give up, I have better things to do.

Since the user has been blocked once again despite my explanations of its use and after two years of good use, the probe will start sending notifications to multimedia-alerts@lists.wikimedia.org.

I am letting the Multimedia and Contributors-Team deal with this task and the sub task T115469 cause there are some social issues I refuse to be involved in.

I agree with James that we shouldn't be running tests in production.

In T115465#1742290, @hashar wrote:

Surely blocking the user again is a constructive way to prioritize the work that needs to be done on T115469 ? I give up, I have better things to do.

I unblocked the bot, so i am responsible (i am not that one who blocked the bot the first time). The bot can run again when stuff is fixed. I am very sorry... but blocking malfunctioning bots until they are fixed is standard practice.

In T115465#1742290, @hashar wrote:

Surely blocking the user again is a constructive way to prioritize the work that needs to be done on T115469 ? I give up, I have better things to do.

Since the user has been blocked once again despite my explanations of its use and after two years of good use, the probe will start sending notifications to multimedia-alerts@lists.wikimedia.org.

I am letting the Multimedia and Contributors-Team deal with this task and the sub task T115469 cause there are some social issues I refuse to be involved in.

No. I am the Multimedia Team and the Editing Department. I have already told you to switch this off, and that we do not agree with it running. You have refused, claiming that Operations need it to verify that Swift works in production. That makes it your problem, or theirs. Please fix it by stopping this task.

Per @Jdforrester-WMF in -releng: "We decided that we definitely wanted it deleted this morning at the team meeting."

Let's delete it :)

Change 247895 had a related patch set uploaded (by Greg Grossmeier):
Remove the UploadWizard API test from production

https://gerrit.wikimedia.org/r/247895

Change 247895 merged by jenkins-bot:
Remove the UploadWizard API test from production

https://gerrit.wikimedia.org/r/247895