* feat(telemetry): add ability to report deploy method to "echo" via telemetry

* feat(telemetry): code formatting

It looks like gate and deck were overriding `buildServiceSettings` just to avoid calling `setAddress`. But then when the JVM flags got added to that method, they both got skipped. Since `deck` doesn't use a JVM, that's no big deal, but for `gate` it's an issue.

Held my nose and went for a minimally invasive change rather than a more holistic cleanup.

…ker#1549)

* refactor(kubernetes): separate V1 and V2 account validators

* feat(kubernetes): add warning for v1 accounts

* feat(cli): only expose providerVersion for kubernetes accounts

* feat(kubernetes): default new and unspecified accounts to V2

* feat(plugins): lays down plugin-manifest for Deck

* chore(style): spotless apply

* fix(plugins): using template for plugin entries for deck

* fix(plugins): corrected filename for plugin-manifest.json

Co-authored-by: Cameron Motevasselani <cmotevasselani@gmail.com>

1. Add an AbstractEditCiCommand class that could be inherited from. For now only codebuild account
command extends this class. There could be more in the future.
2. Mark assume-role and account-id as not required so that the static credentials could be used directly
3. Add an API in halyard daemon to set custom fields for CI providers

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

)

Co-authored-by: Travis CI User <travis@example.org>

Co-authored-by: Travis CI User <travis@example.org>

Co-authored-by: Travis CI User <travis@example.org>

* chore(containers): tell curl to fail if it gets an HTTP error

This will prevent errors from going unnoticed as happened with the
missing AWS client.

* chore(containers): Upgrade the bundled tools

…1565)

* feat(stats): Rename telemetry to stats, enable by default
* fix(stats): Rename TelemetryWrapper to StatsWrapper

Co-authored-by: Cameron Fieber <cfieber@netflix.com>

Co-authored-by: Travis CI User <travis@example.org>

…aker#1572)

* feat(k8s/deploy): Add custom health check to service-settings.

Need the ability to override the default commands created and populated
in the Deployments readinessprobe. In our use case mTLS has been enabled
and we need a way to modify the readiness probe in each service so that
you can pass in the client certificate.

By adding this functionality Halyard can support a much wider array of potential
configurations for health checks for cases where a simple check of /health might not
be desirable.

* Update var to the correct generic type

* fix(k8s/deploy): Update var to the correct generic type

* chore(fix/style): ran spotlessJava to fix style issues

Co-authored-by: runner <runner@fv-az51>

Co-authored-by: runner <runner@fv-az50>

Co-authored-by: runner <runner@fv-az117.internal.cloudapp.net>

Co-authored-by: runner <runner@fv-az76>

Co-authored-by: runner <runner@fv-az51>

…er#1578)

…pinnaker#1580)

This reverts commit fd2ef27.

Slack is littered with people having this issue. It looks like this was added in halyard 1.23 on 8/27/2019. So anyone with a config after that will have this issue.

…pinnaker#1584)

The mergify config currently doesn't distinguish between merging to master and
merging to a release branch.  It currently never merges to release branches
because mergify doesn't have permission to merge to these branches (due to
branch protection rules).

I don't want to give mergify that permission without updating the config file
here, as then it would merge to release branches after any approval, whereas we
only want to merge to release branches if a PR is approved by a release manager.

This updates the mergify config to scope the existing rules to only apply to
the master branch, and adds a new rule to merge to release branches upon
approval by a release manager.

Because the 1.17 branch (and likely the 1.18 branch in some repos) is still
exclusively using Travis CI instead of github actions, also add a rule the
merges if Travis CI passes to account for these branches. This rule can be
deleted once all of the active branches in the repo are using Github actions.

Co-authored-by: Michael Plump <plumpy@google.com>

Co-authored-by: runner <runner@fv-az33>

Co-authored-by: runner <runner@fv-az51.internal.cloudapp.net>

* fix(plugins): changing plugin-manifest.js to .json file type

* fix(plugins): remove comments from json file, handle multiple plugins

* feat(plugins): add plugins to all services

* refactor(plugins): remove introspection

* refactor(plugins): change Boolean to boolean

Co-authored-by: runner <runner@fv-az118.internal.cloudapp.net>

Co-authored-by: runner <runner@fv-az51>

…#1589)

…innaker#1590)

This change breaks deploying any version of Spinnaker <= 1.18
using Halyard, so is being reverted.

This reverts commit 4ed4d26.

Co-authored-by: runner <runner@fv-az51>

Co-authored-by: runner <runner@fv-az51>

Co-authored-by: runner <runner@fv-az51>

Co-authored-by: runner <runner@fv-az51>

* feat(plugins): add plugins to all services

* fix(regression): update version check to handle non-semver releases

* fix(plugins): do not deploy to kayenta until 1.20.0

removed by spinnaker@43dfc67

In addition to putting back something that was there previously, openssl is in the ubuntu
container, so let's have it here as well.

Co-authored-by: Byron David <david.byron@avast.com>

Co-authored-by: runner <runner@fv-az50>

…ker#1600)

Currently if the Halyard daemon was created by another user, we
kill it and try to start one of our own (likely failing as we
probably don't have permission to kill it).

This is because we look up the PID in the pid file, then try to see
if the process exists with 'ps' but don't pass the '-e' flag to get
all processes (including those created by other users). This is
particularly important given that the Halyard install process specifically
asks you to specify a user with which to run Halyard.

* chore(build): move spinnaker library versions out to gradle.properties

* chore(build): upgrade spinnaker gradle plugin

Whoops, missed this part that used to be in init-publish.gradle.

I'll wire all of these up to the bumpdeps action, but I'd like to test these larger jumps first to make sure they go okay...

Co-authored-by: root <root@7c2d69cf985a>

…r#1606)

…yml (spinnaker#1607)

* chore(gradle): Upgrade gradlew to 5.6.4

* chore(containers): Make Dockerfiles work with buildtool's containers.yml

* chore(release): remove jenkins build scripts

…#1608)

* chore(java11): Compile with Java 11 (but targeting Java 8)

* chore(java11): Fix compilation for developers using a Java 11 JDK

* chore(java11): Use a Java 11 JRE

Co-authored-by: root <root@630f71d41ecb>

Co-authored-by: root <root@9be1526fc31e>

Co-authored-by: root <root@3bbaae2b3f97>

Co-authored-by: root <root@60135c639e6f>

Co-authored-by: root <root@6891b1263a18>

…ce the whole bucket is public (spinnaker#1614)

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

Co-authored-by: root <root@458842d3571f>

Co-authored-by: root <root@2e5c1c42d060>

Co-authored-by: root <root@e7992a1c49ce>

Co-authored-by: root <root@18a36aa4866f>

…pinnaker#1620)

Related to: https://github.com/spinnaker/governance/blob/master/rfc/legacy_artifacts_ui_removal.md

Now that the `artifacts` and `artifactsRewrite` flags are [removed from Deck](spinnaker/deck#8184) in preparation for release 1.20, let's remove their corresponding `hal` commands and provide a message for users deploying >= 1.20 that these fields are no longer necessary.

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

Co-authored-by: root <root@9c300884da21>

Co-authored-by: root <root@607dfd992c8d>

Co-authored-by: root <root@4da054c07e9b>

updates to gradle 6.3

updates github actions workflows for new release flags

Co-authored-by: root <root@27f6a48b6497>

…feature flags (spinnaker#1619)

* feat(cli): remove Travis feature flag

* feat(cli): remove Wercker feature flag

* feat(cli): remove Gremlin feature flag

* feat(cli): remove infrastructureStages feature flag

Co-authored-by: root <root@4d3736916f82>

Co-authored-by: root <root@7cfcb65d82e3>

…pinnaker#1633)

removes explicit version opinion which was causing the build to blow up
in some dependency resolution cases (:halyard-web:depInsight
--configuration runtimeClasspath --dependency kork-core)

also was causing the transitives of kubernetes-client not to end up in
the application/deb

removes unnecessary application of java plugin, we apply java-library to all subprojects
removes unnecessary application of protobuf plugin in halyard-web
removes reference to compile scope, uses implementation scope instead

Co-authored-by: root <root@0e23aba3091b>

Co-authored-by: root <root@35bcda70644a>

…er#1639)

Co-authored-by: root <root@7c412b00885f>

Co-authored-by: root <root@ffb573cc3eb2>

…ker#1640)

* fix(google): Update constructors to account for retry changes

The changes to the retry logic updated some fields from Long to
int (because we won't be waiting 24 days between retries); this
commit just updates the places we called them in Halyard to pass
an int.

* chore(dependencies): Bump front50 version

* feat(telemetry): code formatting

* feat(telemetry): code formatting

* fix(versions): Use thread safe yaml parser (#39)

Co-authored-by: Cristhian Castaneda <cristhian.castaneda@armory.io>

Co-authored-by: root <root@7395642f9bee>

Co-authored-by: root <root@2b09fc3a4455>

…pinnaker#1638)

Co-authored-by: Cameron Motevasselani <cmotevasselani@gmail.com>

Co-authored-by: root <root@94827c35f5fd>

Co-authored-by: root <root@3ba87f91c30c>

…nnaker#1653)

I'm not sure what I was thinking with this original PR; adding the
-e flag will return all processes regardless of PID, and is not
necessary when passing a PID.

This is causing halyard to fail to startup because we think that the
daemon is always running.

This reverts commit 4a686e9.

Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>