* Update to OpenCensus version 0.22.3

This should fix Prometheus error messages on exporting with empty tags
list in Allocator Metrics.

* Updates to the Metrics

Fix tests.

Co-authored-by: Mark Mandel <markmandel@google.com>

that we support.

Writing some more tests for player tracking, saw a pattern in assertions
based on GameServer watch events, so wrapped it in an assertWatchUpdate
and assetNoWatchUpdate.

Grouped this at the bottom of the page with gsToTmpFile as well.

Co-authored-by: Robert Bailey <robertbailey@google.com>

Update the agones-allocator doc to recommend using cert-manager for issuing the service certificate (#1459)

Co-authored-by: Mark Mandel <markmandel@google.com>

I also had to update helm to 2.12 for this to works. I've run e2e and verify that all dashboard where working correctly.
The gameserver dashboard is now using the new gauge graph panel, see screenshot below.

Signed-off-by: Cyril Tovena <cyril.tovena@gmail.com>

Co-authored-by: Mark Mandel <markmandel@google.com>

Add more detaied logging for local SDK server. Make npm and apt-get less noisy. apt-get install openjdk-8-jre spares about 25% of output.

For #1452 .

* SDK conformance tests - enhance logging

Remove noisy npm and apt-get logs.

* Use logrus.Entry to tag all local SDK logs.

* Simple ParseEnvFlags test.

Validate Requests and Limits for each PodTemplate Container. Will be
used by GameServer, Fleet and GameServerset.

Co-authored-by: Mark Mandel <markmandel@google.com>

Add Unit Test to cover the race condition.

Co-authored-by: Mark Mandel <markmandel@google.com>

Tweaked the ordering in the release checklist closer to what
I usually do as well.

Was testing out the node.js example in attempt to replicate #1299, but
needed to make changes to the gameserver.yaml to make sure it worked
correctly.

Corrected spelling and some missleading phrases.

This fixes the break in CI we currently have. We missed changing the
kubectl version in the e2e image when we upgraded to Kubernetes 1.14

Now the 1.13.x version is no longer available on google cloud storage,
so things are breaking.

This should fix that!

Wrote an FAQ for Agones based on user questions that have been received
over time.

Corrected the pass for docker run.

* Use go modules for a website and update go version

`make site-deploy` target is working properly.

* Remove faking the GOPATH step in cloudbuild yaml

Go routine could fire after we run the method that fires
the watched event, so let's wait until we know watching is
actively working.

It's can be useful for libraries that use Agones' code as a library to
interact with Agones, it would be useful to enable all Alpha/Beta
features with one command, rather than having to track all Features that
currently exists.

This provides that functionality.

Co-authored-by: Robert Bailey <robertbailey@google.com>

Local SDK implementations of SetPlayerCapacity and GetPlayerCapacity,
including conformance tests for the Go SDK.

Work on #1033

Add new variable feature_gates to change the Agones installation
parameter.

Co-authored-by: Mark Mandel <markmandel@google.com>

* Added version to stress tests files


fix

* version flag description update

Co-authored-by: Alexander Apalikov <alexander.apalikov@globant.com>

Log error properly when Controller started with wrong parameters.

Co-authored-by: Robert Bailey <robertbailey@google.com>

I got too aggressive with moving logging to Debug, and somehow managed
to pickup the local SDK.

I don't think the local SDK logging should be moved to debug, as the
reason the local SDK server exists is to provide information about what
SDK commands are being run -- so it is an Info level debug operation,

* Alpha GameServer attributes added to SDK

Rather than recreating SDK.GameServer() and SDK.WatchGameServer() in the
alpha SDK, I thought it would be easiest both in the short and long
term to mimic what we have done in the GameServer CRD, and populate the
Alpha attributes from these SDK methods when the feature flag for
Player Tracking is enabled.

This is done by adding logic to the conversion function that converts
the K8s GameServer definition into the SDK GameServer definition.

This includes regenerated gRPC code for the supported clients.

I also had to fix the line endings in the C# gen.sh, so including that
in this PR.

* Update based on review

This PR doesn't complete this functionality, but was enough to warrant
review.

It does two things:
1. The refactoring to move alpha and beta SDK Server implementations
into the same file as the main SDK Server, as being able to reuse the
current logic made sense on review.
2. Initial Implementation of SDK GetPlayerCapacity & SetPlayerCapacity
methods with unit tests. E2E tests can be implemented once #1397 is
resolved.

Next steps will be to implement the local SDK functionality, and
conformance tests for Go for these two functions.

Work on #1033

Added a C# wrapper for the Agones gRPC client built against .NET Standard 2.0 & Code Integrity tests built against .NET Core 2.2.

Commit includes:
 - Core C# Wrapper - Fully async & automatic health checks.
 - Agones CI Gen, Build & Test scripts.
 - Docs.
 - Relevant .gitignore modifications.
 - Make targets for C# SDK
 - Configuration specific auto-generate (DebugProtoGen), for development purposes

Resolves: #884 , #883

Co-authored-by: Mark Mandel <markmandel@google.com>

minor fix

fix wrong example format

* Switch from the normal to module terraform config

For development installation using make target.

* Pass more variables to the module helm.

Refactor terraform.mk calls.

* Remove GKE cluster with a small delay after Helm

Remove Helm Release first and wait for 60 seconds so that forwarding
rule and firewall rules would be cleaned up properly by the GCE controller

* Separate Terraform cluster by name - add variable

Now clusters created using gcloud and terraform would have different
names.

Split terraform destroy to give 60 seconds delay time to clean up the
resources created by `helm_release` Agones.

Co-authored-by: Mark Mandel <markmandel@google.com>

Looks like the rate of html requests to Github has been lowered
which is causing errors with our CI process.

This will retry the test, to resolve rate issues or intermittent
external link issues.

To allow connections from AI bots from any host required a change to the
source code of Super Tux Kart.

This uses the author's fork to allow the connections, and builds an
image from there.

Increments the Super Tux Kart image as well.

* Fix wrong condition check for Memory limit
* Add a function to validate Resource config

Add test function for Agones Controller validate function.
Function ValidateResource() would be reused for Fleet, GSS and GS PodTemplate CPU and Memory Resources validation.