Summary: This adds Phacility "Indigo" as well as uses rgba for colors and hovers improving icon colors.
Test Plan:
Tested each color and hover states. Set test instance to scootaloo.
{F310660}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11836
Summary: Fixes T7325, T7326, T7328. When you have deleted a document already you have to specify content; this makes this more clear to the user in this specific delete pathway. Also, includes bonus bug fix for T7326 where we weren't moving the title of the wiki page with the rest of the page.
Test Plan: moved a wiki doc and verified it had the title I had specified. tried to delete an already deleted doc via setting the content to blank (i.e. hitting save after making some other edits) and got more clear error UI state
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7328, T7325, T7326
Differential Revision: https://secure.phabricator.com/D11829
Summary:
This just cleans things up a little:
- Don't show signature status if the document isn't signable.
- Show "Not Signable" instead of "No One" to make the meaning more clear in this context, where we don't have a "Who should sign:" sort of cue.
Test Plan: {F310538}
Reviewers: btrahan, chad
Reviewed By: chad
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D11834
Summary: This reads better?
Test Plan: reading
Reviewers: btrahan, chad
Reviewed By: chad
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D11833
Summary: Fixes T7335. "help" gets you to a specific diviner doc which is an external link, so make sure the code sets is external for the redirect response in this case.
Test Plan: typed "help" and got some
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7335
Differential Revision: https://secure.phabricator.com/D11830
Summary: This increases the transparent space around the Phabricator logo. The logo itself is the same size. This allows for adding of other logos more easily without needing to alter the space provided. (Like Phacility)
Test Plan:
Reload page, screenshot logo, pull into Photoshop and verify spacing top and bottom.
{F309985}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11828
Summary:
Fixes T7229. Some usability issues around this controller - basically you can't leave comments with it and its not particular useful compared to the revision page.
Ergo, if there is a revision associated with a given diff, just re-direct back to the revision page with the proper diff loaded.
Test Plan: Tried to view a diff on the standalone controller attached to a revision and instead was re-directed to the revision view page with the proper diff loaded.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7229
Differential Revision: https://secure.phabricator.com/D11811
Summary:
Fixes T7298. There are two ways to import a repository that you want to host, today:
- Create it as "hosted", then push everything to it.
- Create it as "imported", let it import, then switch it to "hosted".
- (Neither of these work with SVN.)
We don't specifically recommend one or the other, although I believe both should work, and most users seem to go with the first one.
In the first workflow, the new empty repository imports completely and gets marked "imported", so our default behavior is then to publish commits. This can generate a lot of email/notification/feed spam.
If you're a fancy expert you might turn off "publish" before pushing, but normal users will frequently miss this.
Instead, when we receive an "import-like" push to an empty repository, put the repository back into "importing" after we accept the changes.
This has to be heuristic since we can't know for sure if a push is an import or new commits, but here's a simple rule that should do pretty well. We can refine it if necessary.
Test Plan:
- Created a new empty repository.
- Added some debugging code; verified the "commit count" and "empty" rules were calculated properly.
- Pushed 8+ commits and saw the repo go into "importing", import, and leave "importing".
- Pushed 8+ commits again and saw them publish.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T7298
Differential Revision: https://secure.phabricator.com/D11827
Summary:
Ref T7298. We are currently inconsistent about when we publish feed, email, notifications, audits and Herald rules.
Specifically, there are two settings which impact these things:
- The "importing" flag, which is set when we're importing old commits.
- The "herald-disabled" flag, which was expanded in scope some time ago and now actually means "disable publishing".
Various parts of the pipeline were checking only one of these flags. Instead, all of them should check both.
(For example, we should never email users about importing repositories, nor trigger audits on them.)
Test Plan: See next revision.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T7298
Differential Revision: https://secure.phabricator.com/D11826
Summary:
Ref T6840. This feels a little dirty; open to alternate suggestions.
We currently have a race condition where multiple daemons may load a commit and then save it at the same time, when processing "reverts X" text. Prior to this feature, two daemons would never load a commit at the same time.
The "reverts X" load/save has no effect (doesn't change any object properties), but it will set the state back to the loaded state on save(). This overwrites any flag updates made to the commit in the meantime, and can produce the race in T6840.
In other cases (triggers, harbormaster, repositories) we deal with this kind of problem with "append-only-updates + single-consumer", or a bunch of locking. There isn't really a good place to add a single consumer for commits, since a lot of daemons need to access them. We could move the flags column to a separate table, but this feels pretty complicated. And locking is messy, also mostly because we have so many consumers.
Just exempting this column (which has unusual behavior) from `save()` feels OK-ish? I don't know if we'll have other use cases for this, and I like it even less if we never do, but this patch is pretty small and feels fairly understandable (that said, I also don't like that it can make some properties just silently not update if you aren't on the lookout).
So, this is //a// fix, and feels simplest/least-bad for the moment to me, I thiiink.
Test Plan: Added and executed unit tests.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6840
Differential Revision: https://secure.phabricator.com/D11822
Summary: Fixes T6840. Depends on D11822, which is a little iffy.
Test Plan:
Verified all references to `importStatus` are either:
- SQL patches creating the column;
- reads;
- writes immediately before an insert; or
- explicit updates of the column.
That is, I identified no cases of `setImportStatus(X)->save()` on a Commit which may already exist. This //would// break that.
In general, almost all writes go through `$commit->writeImportStatusFlag()`, which is an explicit update.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6840
Differential Revision: https://secure.phabricator.com/D11823
Summary:
Ref T4340. The attack this prevents is:
- An adversary penetrates your network. They acquire one of two capabilities:
- Your server is either configured to accept both HTTP and HTTPS, and they acquire the capability to observe HTTP traffic.
- Or your server is configured to accept only HTTPS, and they acquire the capability to control DNS or routing. In this case, they start a proxy server to expose your secure service over HTTP.
- They send you a link to `http://secure.service.com` (note HTTP, not HTTPS!)
- You click it since everything looks fine and the domain is correct, not noticing that the "s" is missing.
- They read your traffic.
This is similar to attacks where `https://good.service.com` is proxied to `https://good.sorvace.com` (i.e., a similar looking domain), but can be more dangerous -- for example, the browser will send (non-SSL-only) cookies and the attacker can write cookies.
This header instructs browsers that they can never access the site over HTTP and must always use HTTPS, defusing this class of attack.
Test Plan:
- Configured HTTPS locally.
- Accessed site over HTTP (got application redirect) and HTTPS.
- Enabled HSTS.
- Accessed site over HTTPS (to set HSTS).
- Tore down HTTPS part of the server and tried to load the site over HTTP. Browser refused to load "http://" and automatically tried to load "https://". In another browser which had not received the "HSTS" header, loading over HTTP worked fine.
- Brought the HTTPS server back up, things worked fine.
- Turned off the HSTS config setting.
- Loaded a page (to set HSTS with expires 0, diabling it).
- Tore down the HTTPS part of the server again.
- Tried to load HTTP.
- Now it worked.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T4340
Differential Revision: https://secure.phabricator.com/D11820
Summary:
Fixes T7310. We have a whole mechanism for surfacing update errors, but only surface actual update errors, not pull errors.
Instead, surface pull errors too.
Then format them a little more nicely.
Test Plan: {F309769}
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T7310
Differential Revision: https://secure.phabricator.com/D11821
Summary: Helvetica on modern Windows systems doesn't display very well, bolds are fairly not bold and leading between characters is all over the place. This change picks Segoe as the default Windows font which is easier to read, and the bolds are much crisper. There is even a slight improvement in text density with the improved leading.
Test Plan:
I've been using this locally for a few months, overall it's fairly similiar to Source Sans which people seem to enjoy.
{F309851}
{F309852}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: hach-que, Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11825
Summary: Cleans up spacing, hides footer if nothing present, uses common colors.
Test Plan:
Write some typical for a designer code.
{F309840}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11824
Summary: Uses PHUIObjectBoxView to display lists of diffs in Differential and Diffusion, unless embedded on a dashboard.
Test Plan:
Test Dashboard panel, Differential home, Commit, and Diff
{F282173}
{F282174}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11659
Summary: We had two different blues for links here, cleaning that up.
Test Plan: Read a few Diviner documents
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11818
Summary: Fixes a few issues. The author of the commit is more prominent / not cut off. Auditors is in a more consistent location. More space is available for reasons. Commits by themselves look much less janky. Only downside is actual Audits are now 3 lines vs. 2, but the extra space is used well.
Test Plan:
Test list of audits and commits.
{F309237}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11817
Summary: Third times the charm?
Test Plan: pray
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11816
Summary: Fixes T7319. These need a "%s" for the count where they had a "%d"
Test Plan: plan in D11812 is no longer a lie! (watcher added / removed strings render correctly)
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7319
Differential Revision: https://secure.phabricator.com/D11813
Summary: If we don't have a state in PHUIActionPanelView, don't set the extra padding to display it.
Test Plan: Review in UIExamples.
Reviewers: epriestley, btrahan
Reviewed By: btrahan
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11814
Summary: Fixes T7319. ...except I can't get this working in my sandbox? Changes to the translation file don't seem to show up. TEST PLAN IS A LIE
Test Plan: became a watcher, un became a watcher - saw sensical translated strings
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7319
Differential Revision: https://secure.phabricator.com/D11812
Summary: Fixes T7299. Also re-direct the user to the initial request uri if the signature was required.
Test Plan: made a signature required legalpad doc. visit the instance at a specific uri, signed the document, and ended up at that specific uri
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7299
Differential Revision: https://secure.phabricator.com/D11809
Summary:
Fixes the long uptake we saw on `meta.phacility.com`. I regressed this in D11795.
We make three calls to this method, but only one actually consumes the messages. The other two are just checking to see if there are any messages.
Only move the cursor up if we're actually going to process the messages.
Test Plan: Sort of tricky to test convincingly since it's inherently race-prone, but ran `debug pulllocal` and pushed update messages and saw it pick them up.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D11808
Summary:
I am hitting this error when generating Diviner documentation:
```
COMMAND
'/usr/src/phabricator/bin/diviner' atomize --ugly --book $SOME_BOOK --atomizer 'DivinerPHPAtomizer' -- $SOME_PATHS
STDOUT
(empty)
STDERR
[2015-02-18 23:05:01] EXCEPTION: (RuntimeException) Undefined variable: type at [<phutil>/src/error/PhutilErrorHandler.php:210]
#0 PhutilErrorHandler::handleError(integer, string, string, integer, array) called at [<phabricator>/src/applications/diviner/atomizer/DivinerPHPAtomizer.php:315]
#1 DivinerPHPAtomizer::parseReturnType(DivinerAtom, XHPASTNode) called at [<phabricator>/src/applications/diviner/atomizer/DivinerPHPAtomizer.php:116]
#2 DivinerPHPAtomizer::executeAtomize(string, string) called at [<phabricator>/src/applications/diviner/atomizer/DivinerAtomizer.php:23]
#3 DivinerAtomizer::atomize(string, string, array) called at [<phabricator>/src/applications/diviner/workflow/DivinerAtomizeWorkflow.php:109]
#4 DivinerAtomizeWorkflow::execute(PhutilArgumentParser) called at [<phutil>/src/parser/argument/PhutilArgumentParser.php:396]
#5 PhutilArgumentParser::parseWorkflowsFull(array) called at [<phutil>/src/parser/argument/PhutilArgumentParser.php:292]
#6 PhutilArgument... (87 more bytes) ... at [<phutil>/src/future/exec/ExecFuture.php:416]
#0 ExecFuture::resolvex(NULL) called at [<phutil>/src/future/exec/ExecFuture.php:438]
#1 ExecFuture::resolveJSON() called at [<phabricator>/src/applications/diviner/workflow/DivinerGenerateWorkflow.php:349]
#2 DivinerGenerateWorkflow::resolveAtomizerFutures(array, array) called at [<phabricator>/src/applications/diviner/workflow/DivinerGenerateWorkflow.php:209]
#3 DivinerGenerateWorkflow::buildAtomCache() called at [<phabricator>/src/applications/diviner/workflow/DivinerGenerateWorkflow.php:170]
#4 DivinerGenerateWorkflow::generateBook(string, PhutilArgumentParser) called at [<phabricator>/src/applications/diviner/workflow/DivinerGenerateWorkflow.php:74]
#5 DivinerGenerateWorkflow::execute(PhutilArgumentParser) called at [<phutil>/src/parser/argument/PhutilArgumentParser.php:396]
#6 PhutilArgumentParser::parseWorkflowsFull(array) called at [<phutil>/src/parser/argument/PhutilArgumentParser.php:292]
#7 PhutilArgumentParser::parseWorkflows(array) called at [<phabricator>/scripts/diviner/diviner.php:21]
```
Test Plan: N/A
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11807
Summary: Use modern components, pht
Test Plan: I have no data locally, expect @epriestley to commandeer
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11805
Summary: Ref T7202.
Test Plan: Visited edit subscription page and it worked. Clicked edit link from subscription view page and got to the right place.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7202
Differential Revision: https://secure.phabricator.com/D11803
Summary: This change wraps the icon inline with the text, so smaller width icons have equal spacing between the border and text.
Test Plan:
review a number of different tag with icons, also UIExamples.
{F309048}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11802
Summary: Fixes T7317, allows public to be set on this list controller.
Test Plan: Tested a list of subscribers on a logged in and logged out Diff.
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7317
Differential Revision: https://secure.phabricator.com/D11801
Summary:
Ref T6941. In the cluster (and in other reasonable setups) we've separated SSH load balancers from HTTP load balancers.
In particular, ELBs will not let you load balance port 22, so this is likely a reasonable/common issue in larger clusters in AWS.
Allow users to specify an alternate host for SSH traffic.
Test Plan: Set host to someting different, saw it reflected in UI.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6941
Differential Revision: https://secure.phabricator.com/D11800
Summary: Pretty basic, but you can now search panels by type (query, text, tab).
Test Plan: Searched for a few different types of panels, results look correct
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D11782
Summary: Sets everything up flush, I think this is the only indented block left.
Test Plan:
Write some code.
{F289438}
Reviewers: btrahan, epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley, #design
Differential Revision: https://secure.phabricator.com/D11721
Summary: This port is always HTTP, so use HTTP even if users have set the URI to "https".
Test Plan: Launched server and hit status page, status good.
Reviewers: btrahan, chad
Reviewed By: chad
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D11799
Summary: Fixes T7106. If you have bad credentials AND you've pushed an "update this repository" message into the queue, the loop above this level ends up resetting the timer every time we go through it, so the daemon spins in a loop failing forever.
Test Plan:
- Created a repo with bad credentials.
- Clicekd "updated now" to queue an update message.
- Saw daemon run in a loop.
- Applied patch, no loop.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T7106
Differential Revision: https://secure.phabricator.com/D11795
Summary: Fixes T7284. We were initialized the project name to the empty string, which was making things work like a rename, including automagically adding the old slug.
Test Plan: made a project and no more "empty" tag being made. also don't have that bad transaction story anymore.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7284
Differential Revision: https://secure.phabricator.com/D11794
Summary: Even if you --force, we can't kill PID 0. This sends the process itself the signal, and terminates it.
Test Plan: See D11786.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D11787
Summary: Fixes T7263. Last bit there was to upgrade this dialogue to let users know they are letting their primary email address be exposed in these flows. Depends on D11791, D11792, at least in terms of being accurate to the user as the code ended up strangely decoupled.
Test Plan: wordsmithin'
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7263
Differential Revision: https://secure.phabricator.com/D11793
Summary: Ref T7263. We need this in the oauth case and otherwise it makes sense to include.
Test Plan: used the conduit console and saw my email address included in the results!
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7263
Differential Revision: https://secure.phabricator.com/D11791
Summary: Ref T7123. Turns out that we might throw ConduitClientException now in proxied scenarios. For all but one callsite remove the try / catch bit and don't issue the call for SVN. For the remaining callsite, also don't issue the call for SVN but keep in the exception logic since its renders a pretty error message in the non-proxied case?
Test Plan: played around with diffusion and things looked okay.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7123
Differential Revision: https://secure.phabricator.com/D11789
Summary: Fixes T7256.
Test Plan: Looked at rXPRF0a7a5f69f5d7 in a local instance. things looked great both pre and post patch.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7256
Differential Revision: https://secure.phabricator.com/D11790
Summary: Fixes T7294. This lets legalpad store other documents that don't need signatures but conceptually belong in legalpad.
Test Plan: made a document with signature type "no one" and it saved. viewed the document and noted no signing UI was present.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7294
Differential Revision: https://secure.phabricator.com/D11788
Summary:
In the cluster, the box has a ton of stuff that "looks like a daemon" beacuse it is some other instance's daemon.
Stop `phd restart` from complaining about this if given a "--gently" flag, which is like the opposite of "--force".
(I'll make it `stop --force` at the beginning of a whole-box restart to kill stragglers.)
Test Plan: Ran `bin/phd restart --gently`, etc.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D11784
Summary:
Fixes T7291. There are a class of spam/annoyance attacks here that we should be more strict about preventing, since you can add an individual's address as a mailing list.
This application is likely on the way out so I didn't bother trying to do per-object policies.
Test Plan: Set policy restrictively and could no longer create or edit mailing lists.
Reviewers: joshuaspence, btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T7291
Differential Revision: https://secure.phabricator.com/D11783
Summary: Fixes T7295. Humbling debugging experience but I got it.
Test Plan: saved a legalpad doc without edits over and over and saw no "requires signature" transaction. toggled "requires signature", saved, and saw the transaction.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7295
Differential Revision: https://secure.phabricator.com/D11785
Summary: Fixes T7252. The UI is slightly different than in Maniphest - in Maniphest the error shows up at the bottom and here it shows up the top - but I think the UI here makes sense as you see the error right away on the newly returned dialogue?
Test Plan: set "created after" to "assdaasds" and got an error back. set filter to something that should work and it worked
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T7252
Differential Revision: https://secure.phabricator.com/D11760
Summary:
Fixes T7130. Fixes T7041. Fixes T7012.
Major change here is partitioning clients. In the Phacility cluster, being able to get a huge pile of instances on a single server -- without needing to run a process per instance -- is desirable.
To accomplish this, just bucket clients by the path they connect with. This will let us set client URIs to `/instancename/` and then route connections to a small set of servers. This degrades cleanly in the common case and has no effect on installs which don't do instancing.
Also fix two unrelated issues:
- Fix the timeouts, which were incorrectly initializing in `open()` (which is called during reconnect, causing them to reset every time). Instead, initialize in the constructor. Cap timeout at 5 minutes.
- Probably fix subscriptions, which were using a property with an object definition. Since this is by-ref, all concrete instances of the object share the same property, so all users would be subscribed to everything. Probably.
Test Plan:
- Hit notification status page, saw version bump and instance/path name.
- Saw instance/path name in client and server logs.
- Stopped server, saw reconnects after 2, 4, 16, ... seconds.
- Sent test notification; received test notification.
- Didn't explicitly test the subscription thing but it should be obvious by looking at `/notification/status/` shortly after a push.
Reviewers: joshuaspence, btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T7041, T7012, T7130
Differential Revision: https://secure.phabricator.com/D11769
Summary: Fixes T7275. This makes the error stuff a little more consistent with other modern UIs.
Test Plan: {F307286}
Reviewers: chad, btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T7275
Differential Revision: https://secure.phabricator.com/D11778
Summary:
Right now, taskmasters on empty queues sleep for 30 seconds. With a default setup (4 taskmasters), this averages out to 7.5 seconds between the time you do anything that queues something and the time that the taskmasters start work on it.
On instances, which currently launch a smaller number of taskmasters, this wait is even longer.
Instead, sleep for the number of seconds that there are taskmasters, with a random offset. This makes the average wait to start a task from an empty queue 1 second, and the average maximum load of an empty queue also one query per second.
On loaded instances this doesn't matter, but this should dramatically improve behavior for less-loaded instances without any real tradeoffs.
Test Plan: Started several taskmasters, saw them jitter out of sync and then use short sleeps to give an empty queue about a 1s delay.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D11772
