This is also prrrrobably an awful idea, but in theory we could put, say, HAProxy on each host in front of Apache and have it queue connections while Apache was restarting. That feels like we're building a really flimsy house of cards, but it's something computers could technically do.

Cool. Harbormaster integration will also give you other UI hints like these, to help communicate build status to reviewers so they don't accept a change which has failed or in-progress builds:

This isn't the end of the world, but it's pretty bad if you have a mixture of private and less-private threads (sidebar state and main panel state are inconsistent).

If you integrate with Harbormaster, users will be prompted explicitly when they try to arc land a revision with build failures:

Can you provide some more context about why engineers are hitting the "builds failed" prompt, and continuing through it anyway with failing tests and/or nonbuilding code?

The current ruleset implements all those rules correctly, except that it 200's directly for /status/.

redirect to Javelin

On secure, two not-so-great fixes are:

Sounds good.

Oh, maybe not. Things are kind of weird on secure. We should probably sort that out and get that hitting "real" /status/ somehow.

Oh, hrrm. Maybe that RedirectMatch is actually getting hit in practice by the LB, that's suuuper old. I get this from the host itself:

The actual check is elsewhere in that file:

secure is hitting the magic override code in the second link since we deploy rSERVICES there too.

Yeah, I don't like trying to mess with the LBs very much either. I recall the FB F5's (I think?) took like 75 seconds per API call and you needed to make 12 API calls to remove a host, I just figured AWS LBs might be some kind of special magic LBs that can add and remove elements from a list in about a second.

On health-check, vanilla Phabricator runs this:

I'm pretty sure there's a reasonable way to ask apache to gracefully stop in a blocking way

Maybe "duration of all file upload requests" / "duration of all requests". That isn't a perfect metric but should be a reasonable-ish proxy for real resources we care about, I think.

nice progress bar

I suspect this isn't worth the complexity. If you drag-and-drop, the maximum request size is 4MB (one chunk) and we get a nice progress bar, resumable uploads, and the bucket can remain completely private. And we can do encryption and deduplicate file blocks.

I patched this and deployed it. I'll follow up on the internal report with details.

Thanks, this is fallout from an incident last night. The particulars involved instance and user details so the discussion was kept private. I'll fix this shortly.

The only behavior of --lintall is to make warnings act like errors. Just raise everything at error severity to achieve the same thing without patching anything.

You are probably raising messages at "warning" severity, but want to raise them at "error" severity. See, e.g., here for a description:

@chad, a lot of this is probably ultimately design -- how much passion do you feel for hovercards? The infrastructure is in good shape and I can add the "context hint" piece pretty easily if we want to look at that. Not sure if this is a bigger/more complicated project that you don't want to touch for the moment or whatever, though.

(I assumed it was a keydown vs keypress vs whatever event sort of thing but haven't actually looked into it.)

Yeah, this bites me half the time too.

One spanner in the works here is that not all applications with mail preferences are on Modular Transactions yet. I think these are missing:

We don't plan to implement this.

Allowing users to select a subform from "Create Subtask" isn't technically very difficult, but it starts to potentially make the UI pretty messy and complicated. Some possible implementations are:

For my own reference, T12570 isn't exactly related but can probably be tested/fixed at the same time.

Stealing this since I think it's session/auth related, not profile picture related: I think the user has a legitimate picture, they just aren't allowed to download it since they have a partial session.

T8918 is related.

I'm going to merge this into T12374.

@amckinley Let me know if you need more to go on than the description above or get stuck anywhere.

We'd also have to add locking to prevent a race like this:

My gut is to wait for users to hit those cases before building solutions for them since I think it's unlikely any of that will ever turn into a real problem.

Bug reports require reproduction steps. See Contributing Bug Reports and Providing Reproduction Steps.

See T12314#214002.

Conpherence is repainting after D17713 but notification counts aren't, yet.

This could be prettier, but it's functional and has buttons you can click.

For the migration, let me just publish "run this command if you want to search existing files" in the change log for now, and then we can do the real migration if we see confusion about it. I bet 99% of the time you want to search for something you recently uploaded so the index will be functionally useful in a week even if you miss the guidance in the changelog.

To make the indexing work for new stuff, I think you can do this:

I am just intrigued that user-scalable=no actually means "somewhat user-scalable".

No, I hate all zooming.

(I am glad to see the auto-zoom-on-controls gone and don't care about any other zoom feature.)

Oh -- but double-tap-to-zoom doesn't? 😕

(Specifically, D17714 added user-scalable=no, which probably does the opposite of what this requested.)

Hmm ok. I think that I'm not usually listed explicitly as a reviewer, which means that Reviewers: #some_project_that_i_am_a_member_of doesn't include me. Would you be in favor of adding a new [Differential was reviewed by] field to Herald?

I don't think this behavior has changed. This has always meant "reviewers", not "reviewers who accepted the revision".