Summary:
Ref T12173.
- If we want to fetch a tag, Buildkite needs it as a "branch" (this means more like "ref to fetch").
- The API gets upset if we pass "refs/tags/...", so just pass the tag name without the prefix, which works.
- Do a better job with commits and pass a real branch to fetch.
Test Plan:
- Built a commit with Buildkite.
- Build a revision with Buildkite.
Reviewers: chad
Reviewed By: chad
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T12173
Differential Revision: https://secure.phabricator.com/D17282
Summary: Ref T12173. This might need some additional work but the basics seem like they're in good shape.
Test Plan:
- Buildkite is "bring your own hardware", so you need to launch a host to test anything.
- Launched a host in AWS.
- Configured Buildkite to use that host to run builds.
- Added a Buildkite build step to a new Harbormaster build plan.
- Used `bin/harbormaster build ...` to run the plan.
- Saw buildkite execute builds and report status back to Harbormaster
{F2553076}
{F2553077}
Reviewers: chad
Reviewed By: chad
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T12173
Differential Revision: https://secure.phabricator.com/D17270
Summary: Fixes T11910. I spent a couple of minutes looking for the root cause without much luck, but this will all be obsoleted by an eventual upgrade to `EditEngine` anyway.
Test Plan: Set and unset "Wait for Message", which now worked.
Reviewers: chad, avivey
Reviewed By: avivey
Subscribers: PHID-OPKG-gm6ozazyms6q6i22gyam
Maniphest Tasks: T11910
Differential Revision: https://secure.phabricator.com/D16919
Summary:
Ref T11153. If you have a build plan like this:
- Lease machine A.
- Lease machine B.
- Run client-tests on machine A.
- Run server-tests on machine B.
...and we get machine A quickly, then finish the tests, we currently do not release machine A until the whole plan finishes.
In the best case, this wastes resources (something else could be using that machine for a while).
In a worse case, this wastes a lot of resources (if machine B is slow to acquire, or the server tests are much slower than the client tests, machine A will get tied up for a really long time).
In the absolute worst case, this might deadlock things.
Instead, release artifacts as soon as no waiting/running steps take them as inputs. In this case, we'd release machine A as soon as we finished running the client tests.
In the case where machines A and B are resources of the same type, this should prevent deadlocks. In all cases, this should improve build throughput at least somewhat.
Test Plan:
I wrote this build plan which runs a "fast" step (10 seconds) and a "slow" step (120 seconds):
{F1691190}
Before the patch, running this build plan held the lease on the "fast" machine for the full 120 seconds, then released both leases at the same time at the very end.
After this patch, I ran this plan and observed the "fast" lease get released after 10 seconds, while the "slow" lease was held for the full 120.
(Also added some `var_dump()` into things to sanity check the logic; it appeared correct.)
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T11153
Differential Revision: https://secure.phabricator.com/D16145
Summary:
Ref T11137. This class is removed in D16099. Depends on D16099.
`PhutilURI` now attempts to "just work" with Git-style URIs, so at least in theory we can just delete all of this code and pretend it does not exist.
(I've left "Display URI" and "Effective URI" as distinct, at least for now, because I think the distinction may be relevant in the future even though it isn't right now, and to keep this diff small, although I may go remove one after I think about this for a bit.)
Test Plan:
- Created a new Git repository with a Git URI.
- Pulled/updated it, which now works correctly and should resolve the original issue in T11137.
- Verified that daemons now align the origin to a Git-style URI with a relative path, which should resolve the original issue in T11004.
- Grepped for `PhutilGitURI`.
- Also grepped in `arcanist/`, but found no matches, so no patch for that.
- Checked display/conduit URIs.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T11137
Differential Revision: https://secure.phabricator.com/D16100
Summary: Ref T9456. This makes everything work, except that CircleCI doesn't fetch tags which are not ancestors of branch heads.
Test Plan: Ran passing builds through CircleCI.
Reviewers: chad
Reviewed By: chad
Subscribers: dpaola2, JustinTulloss
Maniphest Tasks: T9456
Differential Revision: https://secure.phabricator.com/D14288
Summary: Ref T9456. Some rough edges and we can't complete the build yet since I haven't written a webhook, but this mostly seems to be working.
Test Plan:
- Ran this build on some stuff.
- Ran a normal HTTP step build to make sure I didn't break that.
{F880301}
{F880302}
{F880303}
Reviewers: chad
Reviewed By: chad
Subscribers: JustinTulloss, joshma
Maniphest Tasks: T9456
Differential Revision: https://secure.phabricator.com/D14286
Summary:
Ref T5822. This prepares for inline compression and garbage collection of build logs.
This reduces the API surface area and removes a log from the "wait" step that would just log a message every 15 seconds. (If this is actually useful, I think we find a better way to communicate it.)
Test Plan:
Ran a build, saw a log:
{F1136691}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T5822
Differential Revision: https://secure.phabricator.com/D15371
Summary: Fixes T10458. These steps are obsolete and have not worked since the last updates to Drydock. They may eventually return in some form, but get rid of them for now since they're confusing.
Test Plan:
- Created a build plan with these steps.
- Removed these steps.
- Verified the build plan showed that the steps were invalid, and that I could delete them.
- Deleted them.
- Added new steps, no obsolete steps were available for selection.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T10458
Differential Revision: https://secure.phabricator.com/D15352
Summary: Ref T4245. These mostly relate to building URIs.
Test Plan: Tried to hunt down as many of these in the UI as I could. Some are a bit tricky but they should be low-risk.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4245
Differential Revision: https://secure.phabricator.com/D14933
Summary:
Fixes T9669. Two issues:
- We were using `repositoryPHIDs` instead of `blueprintPHIDs` for the list of allowed blueprints. Use the correct value.
- We weren't enforcing `allowedBlueprintPHIDs` fully correctly. We //did// require an authorization, so the net effect was correct in nearly all cases, but we could have selected from too large a pool in the case where the application itself was doing the authorization (e.g., from the command line).
Test Plan: Ran a build through Drydock/Harbormaster locally.
Reviewers: chad, tycho.tatitscheff
Reviewed By: chad, tycho.tatitscheff
Subscribers: tycho.tatitscheff
Maniphest Tasks: T9669
Differential Revision: https://secure.phabricator.com/D14368
Summary: Fixes T9631. Build steps created before I added this option may not have it specified, which could throw later. Make handling a little more robust.
Test Plan: Will ask @yelirekim to report back.
Reviewers: chad
Reviewed By: chad
Subscribers: yelirekim
Maniphest Tasks: T9631
Differential Revision: https://secure.phabricator.com/D14336
Summary:
Ref T9519. When acquiring leases on resources:
- Only consider resources created by authorized blueprints.
- Only consider authorized blueprints when creating new resources.
- Fail with a tailored error if no blueprints are allowed.
- Fail with a tailored error if missing authorizations are causing acquisition failure.
One somewhat-substantial issue with this is that it's pretty hard to figure out from the Harbormaster side. Specifically, the Build step UI does not show field value anywhere, so the presence of unapproved blueprints is not communicated. This is much more clear in Drydock. I'll plan to address this in future changes to Harbormaster, since there are other related/similar issues anyway.
Test Plan: {F872527}
Reviewers: hach-que, chad
Reviewed By: chad
Maniphest Tasks: T9519
Differential Revision: https://secure.phabricator.com/D14254
Summary: Ref T9252. Move these into a new "Drydock" group.
Test Plan: Clicked "Add Build Step", saw Drydock steps in a Drydock group.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9252
Differential Revision: https://secure.phabricator.com/D14237
Summary:
Ref T9252. Currently, Harbormaster and Drydock work like this in some cases:
# Queue a lease for activation.
# Then, a little later, save the lease PHID somewhere.
# When the target/resource is destroyed, destroy the lease.
However, something can happen between (1) and (2). In Drydock this window is very short and the "something" would have to be a lighting strike or something similar, but in Harbormaster we wait until the resource activates to do (2) so the window can be many minutes long. In particular, a user can use "Abort Build" during those many minutes.
If they do, the target is destroyed but it doesn't yet have a record of the artifact, so the artifact isn't cleaned up.
Make these things work like this instead:
# Create a new lease and pre-generate a PHID for it.
# Save that PHID as something that needs to be cleaned up.
# Queue the lease for activation.
# When the target/resource is destroyed, destroy the lease if it exists.
This makes sure there's no step in the process where we might lose track of a lease/resource.
Also, clean up and standardize some other stuff I hit.
Test Plan:
- Stopped daemons.
- Restarted a build in Harbormaster.
- Stepped through the build one stage at a time using `bin/worker execute ...`.
- After the lease was queued, but before it activated, aborted the build.
- Processed the Harbormaster side of things only.
- Saw the lease get destroyed properly.
Reviewers: chad, hach-que
Reviewed By: hach-que
Maniphest Tasks: T9252
Differential Revision: https://secure.phabricator.com/D14234
Summary: Ref T9478. This should probably be configurable eventually, but for now treat any 200-block status as success. Also show the result code.
Test Plan:
- Hit a bad URI, saw "HTTP 503" + failure.
- Hit a good URI, saw "HTTP 200" + success.
Reviewers: chad, hach-que
Reviewed By: chad, hach-que
Maniphest Tasks: T9478
Differential Revision: https://secure.phabricator.com/D14226
Summary: Ref T9252. This variable was always wrong but we fell back to just resetting to `HEAD` before. Use the correct variable name.
Test Plan: Verified variable name.
Reviewers: chad, hach-que
Reviewed By: hach-que
Maniphest Tasks: T9252
Differential Revision: https://secure.phabricator.com/D14224
Summary: Ref T9352. See D13635. Build targets can have variables already, but let builds have them too. This mostly enables future use cases (sub-builds, more sophisticated build triggers).
Test Plan: With a custom Herald rule + action like the one in T9352, updated a revision and saw it generate multiple builds with varying parameters.
Reviewers: chad, hach-que
Reviewed By: hach-que
Maniphest Tasks: T9352
Differential Revision: https://secure.phabricator.com/D14222
Summary:
Ref T9252. This primarily allows Harbormaster to request (and Drydock to fulfill) working copies with a patch from a staging area. Doing this means we can do builds on in-review changes from `arc diff`.
This is a little cobbled-together but should basically work.
Also fix some other issues:
- Yielded, awakend workers are fine to update but could complain.
- We can't log slot lock failures to resources if we don't end up saving them.
- Killing the transaction would wipe out the log.
- Fix some TODOs, etc.
Test Plan: Ran Harbormaster builds on a local revision.
Reviewers: hach-que, chad
Reviewed By: chad
Maniphest Tasks: T9252
Differential Revision: https://secure.phabricator.com/D14214
Summary:
Ref T9123. To run upstream builds in Harbormaster/Drydock, we need to be able to check out `libphutil`, `arcanist` and `phabricator` next to one another.
This adds an "Also Clone: ..." field to Harbormaster working copy build steps so I can type all three repos into it and get a proper clone with everything we need.
This is somewhat upstream-centric and a bit narrow, but I don't think it's totally unreasonable, and most of the underlying stuff is relatively general.
This adds some more typechecking and improves data/type handling for custom fields, too. In particular, it prevents users from entering an invalid/restricted value in a field (for example, you can't "Also Clone" a repository you don't have permission to see).
Test Plan: Restarted build, got a Drydock resource with multiple repositories in it.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9123
Differential Revision: https://secure.phabricator.com/D14183
Summary:
Ref T9252. For building Phabricator itself, we need to have `libphutil/`, `arcanist/` and `phabricator/` next to one another on disk.
Expand the Drydock WorkingCopy resource so that it can have multiple repositories if the caller needs them.
I'm not sure if I'm going to put the actual config for this in Harbormaster or Drydock yet, but the WorkingCopy resource itself should work the same way in either case.
Test Plan: Restarted a Harbormaster build which leases a working copy, saw it build as expected.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9252
Differential Revision: https://secure.phabricator.com/D14180
Summary:
Ref T9252. Currently, Harbormaster does this when trying to acquire a working copy:
- Ask for a working copy.
- Yield for 15 seconds.
- Check if we have a working copy yet.
That's OK, but Drydock takes ~1s to acquire a working copy lease if a resource is already available, so we end up doing this:
- T+0: Ask for a working copy.
- T+0: Yield for 15 seconds.
- T+1: Working copy lease activates.
- T+15: Working copy lease is used.
- T+16: Build finishes.
So we end up spending about 2 seconds doing work and 14 seconds sleeping.
One way to fix this would be to fiddle with the yield duration, so we yield for 1, 2, 4, ... seconds or something. This probably isn't a bad idea for longer leases (i.e., wait for 15, 30, 45 ... seconds or similar) but it implies a lot of churn for short leases.
Instead, let tasks "awaken" other tasks when they complete. The "awaken" operation means: if a task is in a yielded state (no failures, no owner, explicitly yielded, future expires time), pretend it only yielded until right now instead of whenever it really yielded to.
Basically, this rewrites history so that even though Harbormaster did a `yield(15)`, we pretend it did a `yield(4)` after we activate the lease if lease activation took 4 seconds.
If this misses, it's fine: we fall back to the normal yield behavior and things move forward normally a few seconds later.
If it hits, we get a more nimble process pretty cleanly.
Test Plan:
- Restarted a build plan (lease working copy + run `ls`) with this patch no-op'd, took about 16 seconds.
- Restarted a build plan with this patch active, took about 1 second.
Reviewers: hach-que, chad
Reviewed By: chad
Maniphest Tasks: T9252
Differential Revision: https://secure.phabricator.com/D14178
Summary: Ref T9252. This mostly cleans up future and log handling, and edges us closer to being able to do useful work with Harbormaster / Drydock.
Test Plan:
- Added a "Run `ls -alh`" step to my trivial build plan.
- Ran it a bunch of times.
- Worked great.
- Also did an HTTP plan.
{F835227}
{F835228}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9252
Differential Revision: https://secure.phabricator.com/D14161
Summary: Ref T9252. This is still crude in a few ways but basically works, at least for commits.
Test Plan:
- Made a build plan with just this build step.
- Ran `bin/harbormaster build --plan 10 ...` on a commit.
- It actually built a working copy, leased it, took no action, and released the lease. MAGIC~~~
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9252
Differential Revision: https://secure.phabricator.com/D14160
Summary: Ref T1049. This ensures the Harbormaster build target is associated with leases, so in the future we can query things and find out whether builds are still running with associated leases.
Test Plan: Leased a host, checked the DB and saw the field populated.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: joshuaspence, Korvin, epriestley
Maniphest Tasks: T1049
Differential Revision: https://secure.phabricator.com/D10870
Summary: Ref T9205. This is a likely fix.
Test Plan: This isn't straightforward to test in the upstream unless you have custom code on top of it.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T9205
Differential Revision: https://secure.phabricator.com/D13928
Summary:
Ref T8659. In the general case, this eventually allows build processes to do things like:
- Upload build results (like a ".app" or ".exe" or other binary).
- Pass complex results between build steps (e.g., build step A does something hard and build step B uses it to do something else).
Today, we're a long way away from having the infrastructure for that. However, it is useful to let third party build processes (like Jenkins) upload URIs that link back to the external build results.
This adds `harbormaster.createartifact` so they can do that. The only useful thing to do with this method today is have your Jenkins build do this:
params = array(
"uri": "https://jenkins.mycompany.com/build/23923/details/",
"name": "View Build Results in Jenkins",
"ui.external": true,
);
harbormaster.createartifact(target, 'uri', params);
Then (after the next diff) we'll show a link in Differential and a prominent link in Harbormaster. I didn't actually do the UI stuff in this diff since it's already pretty big.
This change moves a lot of code around, too:
- Adds PHIDs to artifacts.
- It modularizes build artifact types (currently "file", "host" and "URI").
- It formalizes build artifact parameters and construction:
- This lets me generate usable documentation about how to create artifacts.
- This prevents users from doing dangerous or policy-violating things.
- It does some other general modernization.
Test Plan:
{F715633}
{F715634}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T8659
Differential Revision: https://secure.phabricator.com/D13900
Summary:
Ref T8089. We have a lot of broken/confusing/prototype build steps that I want to hide from users when we unprototype Harbormaster.
The dialog is also just kind of unwieldy.
Organize this UI a little better and put all the sketchy junk in a "prototypes" group that you can't see unless prototypes are enabled.
This doesn't break anything (the old steps will still work fine), but should reduce user confusion.
Test Plan:
Old UI:
{F691439}
New UI (prototypes off):
{F691440}
New UI (prototypes on):
{F691441}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T8089
Differential Revision: https://secure.phabricator.com/D13803
Summary: Use `PhutilClassMapQuery` where appropriate.
Test Plan: Browsed around the UI to verify things seemed somewhat working.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D13429
Summary:
Ref T8095. Two general problems:
- I want Harbormaster to own all lint and unit test results.
- I don't want users to have to configure anything for `arc` to keep working automatically.
These are in conflict because generic lint/unit test ownership in Harbormaster requires that build targets exist which we can attach build results to. However, we can't currently create build targets on demand: Harbormaster assumes it is responsible for creating targets, then running code or making third-party service calls to actually run the builds.
I considered two broad approaches to let `arc` push results into Harbormaster without requiring administrators to configure some kind of "arc results" build plan:
# Add magic target PHIDs like `PHID-MAGIC-this-is-really-arc-unit`.
# Add new code to build real targets with real PHIDs.
(1) is probably a bit less work to get off the ground, but I think it's worse overall and very likely to create more problems in the long run. I particularly worry that it will lead to a small amount of special casing in a very large number of places, which seems more fragile.
(2) is more work upfront but I think does a better job of putting all the special casing in one place that we can, e.g., more reasonably unit test, and letting the rest of the code rarely/never care about this case since it's just dealing with normal plans/steps/targets as far as it can tell.
This diff introduces "autoplans", which are source templates for plans/steps. This let us "push" these targets into Harbormaster. Hypthetically, any process "like" arc can use autoplans to upload test/lint/etc results. In practice, probably only `arc` will ever use this, but I think it's still quite a bit cleaner than the alternative despite all the generality.
Workflow is basically:
- `arc` creates a diff.
- `arc` calls `harbormaster.queryautotargets`, passing the diff PHID and saying "I have some lint and unit results I want to stick on this thing".
- Harbormaster builds the plan, steps, and targets (if any of them don't already exist), and hands back the target PHIDs so `arc` has a completely standard-looking place to put results.
- `arc` uploads the test results to the right targets, as though Harbormaster had asked it to run unit/lint in the first place.
(This doesn't actually do any of that yet, just sets things up.)
I'll maybe doc turn that ^^^^^^ into a doc for posterity since I think it's hard to guess what an "autotarget" is, but I'm going to grab some lunch first.
Test Plan:
- Added unit tests to make sure we can build these things properly.
- Used `harbormaster.queryautotargets` to build autotargets for a bunch of diffs.
- Verified targets come up in "waiting for message" state.
- Verified plans and steps are not editable.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: hach-que, epriestley
Maniphest Tasks: T8095
Differential Revision: https://secure.phabricator.com/D13345
Summary: All classes should extend from some other class. See D13275 for some explanation.
Test Plan: `arc unit`
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D13283
Summary: Use `__CLASS__` instead of hard-coding class names. Depends on D12605.
Test Plan: Eyeball it.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: hach-que, Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D12806
Summary:
Resolves T5987. This build step was at some point converted to use yielding, which meant that whenever the build step executes it will create a new log. This checks to see if there is an existing log before creating a new one and uses that instead.
Long term we're going to need some way of attaching data to `PhabricatorWorkerYieldException` that can be read when the build step starts again; this will allow us to move more build steps off `while (...) { ... sleep(X); }` loops and onto yielding.
Test Plan: Tested locally.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley
Maniphest Tasks: T5987
Differential Revision: https://secure.phabricator.com/D10383
Summary: Ref T5936. This implements build implementations aborting early when the build has since been restarted. Build steps now periodically poll to see if the build's current generation does not match their generation, and they throw a `HarbormasterBuildAbortedException` if that is the case.
Test Plan: Tested locally on my machine with the sleep build step.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Maniphest Tasks: T5936
Differential Revision: https://secure.phabricator.com/D10322
Summary:
Resolves T5831. This modifies the Drydock SSH interface to execute commands under Powershell when the target host platform is Windows. Powershell is far more featured than cmd.exe, and more closely resembles a UNIX shell.
Currently Powershell outputs stderr as an XML blob on a line, and while this code currently doesn't use that, it will allow us in the future (planned next week) to redirect that output to the stderr log instead of having it all merged in with stdout under cmd (where there is no way to distinguish it).
Test Plan:
Ran various native commands and PowerShell commands from a Harbormaster build, including things like:
```
Write-Host ("my test" + ${build.id})
```
and saw:
```
my test679
```
in the output.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Maniphest Tasks: T5831
Differential Revision: https://secure.phabricator.com/D10248
Summary: We've received feedback that the "core - exception" is incredibly confusing, to the point where developers see this and write off the build failure as a Phabricator error that is unrelated to their changes.
Test Plan: Ran a build with a `exit 1` run step, didn't see the "core - exception" appear.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10090
Summary: This makes input artifacts imply the appropriate build step dependencies in the build plan. That is, if you use a host artifact in a build step, it will then implicitly depend on the 'Lease Host' step.
Test Plan: Viewed the build plan with the artifacts, saw the dependencies. Ran a build, saw everything execute in the correct order.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10089
Summary:
Depends on D9806. This implements the build simulator, which is used to calculate the order of build steps in the plan editor. This includes a migration script to convert existing plans from sequential based to dependency based, and then drops the sequence column.
Because build plans are now dependency based, the grippable and re-order behaviour has been removed.
Test Plan: Tested the migration, saw the dependencies appear correctly.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D9847
Summary: Ran `arc lint --apply-patches --everything` over rP, mainly to change double quotes to single quotes where appropriate. These changes also validate that the `ArcanistXHPASTLinter::LINT_DOUBLE_QUOTE` rule is working as expected.
Test Plan: Eyeballed it.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin, hach-que
Differential Revision: https://secure.phabricator.com/D9431
Summary: Fixes T5062. See inlines.
Test Plan: Did not test whatsoever.
Reviewers: hach-que
Reviewed By: hach-que
Subscribers: epriestley
Maniphest Tasks: T5062
Differential Revision: https://secure.phabricator.com/D9132
Summary:
For Harbormaster tasks which want to poll or wait, this lets them say "try again a little later" without having to sleep and hold a queue slot.
This is basically the same as failing, except that we don't increment the failure counter. Instead, we just set the current lease to the correct length and then exit. The task will be retried after the lease expires.
Test Plan: Using both `bin/harbormaster` and `phd debug taskmaster`, ran a lot of waiting tasks through the queue, faking them to either yield or not yield in a controlled manner. The queue responded as expected, yielding tasks appropraitely and retrying them later.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D8792
Summary:
This hooks up all the pieces of the build pipeline so `harbormaster.sendmessage` actually works. Particularly:
- Candidate build steps (i.e., those which interact with external systems) can now "Wait for Message". This pauses them indefinitely when they complete, until something calls `harbormaster.sendmessage`.
- After processing a target, we check if we should move it to PASSED or WAITING.
- Before updating a build, we move WAITING targets with pending messages to either PASSED or FAILED.
- I added an explicit "Building" state, which doesn't affect workflows but communicates more information to human users.
A big part of this is avoiding races. I believe we get the correct behavior no matter which order events occur in:
- We update builds after targets complete and after we receive messages, so we're guaranteed to update once both these conditions are true. This means messages can't be lost (even if they arrive before a build completes).
- The minor changes to the build engine logic mean that firing additional build updates is always safe, no matter what the current state of the build is.
- The build itself is protected by a lock in the build engine.
- The target is not covered by an explicit lock, but for all states only the engine (waiting) //or// the worker (all other states) can interact with it. All of the interactions also move the target state forward to the same destination and have no other side effects.
- Messages are only consumed inside the engine lock, so they don't need an explicit lock.
Test Plan:
- Made an HTTP request wait after completion, then ran a pile of builds through it using `bin/harbormaster build` and the web UI.
- Passed and failed message-awaiting builds with `harbormaster.sendmessage`.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley, zeeg
Differential Revision: https://secure.phabricator.com/D8788
