Wed, Nov 15
Tue, Oct 31
Thanks. Okay to close this task? Looks like @Yurik +2'd.
https://gerrit.wikimedia.org/r/#/c/387628 and https://gerrit.wikimedia.org/r/#/c/387629 posted. @Anomie would you please review? Note, I pointed ApiZeroPortal:getCustomPrinter() documentation readers back to the documentation concerning ApiRawJsonPrinter.
Wed, Oct 25
Here's what I see now using that previously erroring request.
Heads up for those following the ticket: @Anomie and I talked and I have this on my list for a follow on patches to clarify more of the why in the comments. Regarding the comment on the class and not the method, I'll fix that - thanks for spotting that @Anomie. As a reminder, my impending time off could arrive any moment - although documentation changes are simple I want to get it right, so please bear with me in case it takes some time...possibly until after returning.
Tue, Oct 24
@Aklapper I'm unsure if it's PdfHandler. @matmarex it has a header match, but I'm not sure - when I spot checked yesterday it seemed like it was mainly PDFs, not images, but to be fair it's hard to say. Looping in @MarkTraceur @MaxSem @Tgr @brion for ping. Anybody have ideas on what might be going on and how to resolve?
Mon, Oct 23
Thanks, @Anomie ! Do https://gerrit.wikimedia.org/r/#/c/386014/ and https://gerrit.wikimedia.org/r/#/c/386012/ capture the sort of detail you had in mind? Thanks for the reassurance about the non-breakage.
Fri, Oct 20
Say, I noticed that calls to, for example, http://php5.local.wmftest.net:8080/w/api.php?action=zeroconfig&type=message&agent=<someUA> on my Vagrant instance with header forging (which would be equivalent to https://en.wikipedia.org/w/api.php?action=zeroconfig&type=message&agent=<someUA> with VCL determining an X-CS header independently or https://en.wikipedia.org/w/api.php?action=zeroconfig&type=message&agent=<someUA>&X-CS=TEST1 if using the parameter debugging technique...which gets send to the inbound origin) exercise ApiZeroBanner and ApiRawJsonPrinter. This type of URL is consumed from an internet context by native apps (and historically the FFOS web app).
Oct 17 2017
Oct 16 2017
^ and @Johan I think you probably know, too.
@Dereckson thanks for pinging on this. @Deskana is now in the product owner role for this technology's rollout. I think there may be a couple outstanding questions on this, and we should defer to him on rollout.
Oct 11 2017
An update: we looked into this. In short, we got stuck at the point of drivers. It appears that there may be a way to pure OpenCL-oriented driver support (i.e., without proprietary driver needs), along the lines of the links here, but it wasn't readily apparent. My impression is AMD may need to be asked how we'd do this. I'm going to be on some family leave for a longer period starting pretty soon (everything's fine, BTW), so I just wanted to note I probably won't have a lot of time to look at this between now and then.
Oct 4 2017
Just so we have it here, for TensorFlow people, there's an encouraging comment at https://github.com/tensorflow/tensorflow/issues/22#issuecomment-334154564 .
+ @Deskana is taking over product owner duties for rollout. There are a few details about near-term and medium term blocking tasks, and I defer to him about those.
Sep 26 2017
Thanks for asking. Aligning resources for code review is on the project manager's task list.
Re-opening, Gergo to implement - this is based on product manager request as a risk mitigation. None of us believe this is particularly hard (although there are probably differences of opinion about YAGNI), although having a configuration variable might at least streamline things in case one wiki is having an issue but all the others are okay.
Is this something we can consult with communities on as we rollout? If no one wants it, we don't have to implement it. If someone wants it, then we can add support for it. What do you think, @Deskana?
Sep 25 2017
Following up on this, I arranged some time with @Ottomata to take a look into this.
Sep 22 2017
@Mrjohncummings I just emailed you with an intro to a potential collaborator. Heads up!
Sep 21 2017
Note to future selves: we'd probably want the filename to contain the human readable File: name so people and machines can generate search results more easily.
Ha! That’s what they all say :P
Thanks, @Ottomata. Any chance you could take a look at the GPU? I’d like to watch to learn something about the setup of this in Debian. We reserved a couple machines for this FY (ETA Q3+4) and may want to add GPUs...the thought being we’d use the same unit.
Sep 20 2017
@Ottomata do you know if it ws an s9150, or was it a W9100? I'm seeing the 04:00.0 Display controller: Advanced Micro Devices, Inc. [AMD/ATI] Hawaii XT GL [FirePro W9100] from lspci and the following from lshw
Sep 14 2017
We cannot expedite requests without @mark specifically overriding this process. In particular on granting a user sudo rights, sorry!
I approve of the access request being fulfilled. I'm in support of expediting if possible.
Sep 8 2017
Sep 6 2017
^ Note, removed inaccurate statement about reaching users...that was for a different task.
This appears to require more technical investigation.
Does anybody have a way to reach Jeroun De Dauw on Phabricator? This looks like a Nuke extension thing and that name is attached per https://www.mediawiki.org/wiki/Developers/Maintainers.
Closing for now. Please re-open if reproducible.
Closing for now as it appears to be working. Please re-open if reproducible.
Aug 29 2017
Yeah, the line should probably say:
Aug 28 2017
@CKoerner_WMF what way would you recommend broaching this with the community members, if any?
Aug 24 2017
Tried with 2.6.201-alpha-2017-08-24 and the issue doesn't seem to be surfacing.
Aug 23 2017
@MarkTraceur, adding for your review as well in case anything pertinent for Multimedia.
Here's my config:
Aug 21 2017
Needs technical investigation.
Pulling to Tracking (reflecting that the content is community provided) and Lowest (to reflect general case of most devices supporting JS; although it's true that there are some non-JS users, particularly if they were switching to desktop mode from a compression proxy browser).
It looks like this may have been a thumbnail race condition, but it doesn't appear to be reproducible. Closing as Invalid on procedural grounds, please re-open with screenshots / animation if able to reproduce.
Sliding over for product owner review when product owner for Multimedia starts.
This is probably pertinent to Structured Data on Commons.
Per @Aklapper guidance, closing out.
This was simply a symptom of the deployment approach and this is a testing server. Future deployments when done according to standard would avoid this issue.
This does in fact appear to be a bug. Moving to the Triaged column.
@dpatrick bringing to your attention in case this is something your team might want to weigh in on.
Aug 18 2017
I think there are a few things being discussed on this ticket. Are people around next week to do a hangout? We should do a review to get everyone up to speed, and then figure out how we can all support each other better in the future around things of this nature.
Aug 17 2017
I don't rightly recall.
Aug 16 2017
Hey all, are we sure this is only (EDIT: logged in) and is the service down? I noticed the button on the Sierra Leone page as an anon and on tap the map load seems to stall out. Here's what I'm seeing.
Aug 11 2017
Aug 8 2017
Aug 7 2017
Thanks, @ABorbaWMF. Say, I noticed if I do the following:
Triaging this morning. For a potential team lead call in the future.
The next action on this would be to conduct further technical investigation and scope the level of effort for potentially bugfixing.
We're triaging bugs this morning. Not pressing at the moment, but noting we've seen it.
We're doing triaging this morning.
Moving this to the "Product owner backlog" for classification.
Presently we're not working on ImageTweaks, but I've modified this task to be part of that project for some future date. This will require further investigation from a UX perspective when that time comes.
Next action would be to conduct technical root cause analysis, then further scope for potential bugfix.
Next step on this would be to further investigate and scope level of effort before committing to work.
Closing task as "declined" until further reproduction disclosed.
@Nirzar what treatment would you suggest?
@Nirzar what treatment do you suggest?
Updating status and priority until further information received.