This configuration allows to use EU based NewRelic account.

Signed-off-by: Syamala Umamaheswaran <shykart2203@gmail.com>

Signed-off-by: Syamala Umamaheswaran <shykart2203@gmail.com>

NewRelicEmitter URL is the complete url and not the base URL as configuredin the NewRelicConfig

Signed-off-by: Syamala Umamaheswaran <shykart2203@gmail.com>

Signed-off-by: Syamala Umamaheswaran <shykart2203@gmail.com>

Add NewRelicEmitter changes to release notes

Signed-off-by: Syamala Umamaheswaran <shykart2203@gmail.com>

* NOTE: we don't log failed type conversions. This is probably something
we want to start doing once we allow users to bypass dex. For now its
probably ok to fail silently because our token structure will be
consistent for the time being.

https://github.com/concourse/concourse/projects/49#card-31357785
https://github.com/concourse/concourse/projects/49#card-36796693

Co-authored-by: Ciro S. Costa <cscosta@pivotal.io>
Signed-off-by: Josh Winters <jwinters@pivotal.io>

Using the limit-active-tasks placement strategy gives the perfect opportunity
to dynamically scale workers based on tasks running. This metric makes
it easier for operators to implement such a scaling.

#5057

Co-authored-by: Torben Neufeldt <torben.neufeldt@oss.volkswagen.com>
Signed-off-by: Hannes Hasselbring <hannes@oss.volkswagen.com>

Co-authored-by: Torben Neufeldt <torben.neufeldt@oss.volkswagen.com>
Co-authored-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>
Signed-off-by: Hannes Hasselbring <hannes@oss.volkswagen.com>

Co-authored-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>
Signed-off-by: Hannes Hasselbring <hannes@oss.volkswagen.com>

Co-authored-by: Hannes Hasselbring <hannes@oss.volkswagen.com>
Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>

- To avoid test pollution all previously registered collectors are unregistered in
  garbage collection test

Co-authored-by: Hannes Hasselbring <hannes@oss.volkswagen.com>
Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>

…to be available

- Make metrics test more stable

Co-authored-by: Hannes Hasselbring <hannes@oss.volkswagen.com>
Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>

- introduce wait group so we know the context is cancelled before running the tests

Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>

- Close sshClient Conn on keepalive -> SendRequest operation error
- Add 5 second timeout for keepalive -> SendRequest operation

Signed-off-by: Sameer Vohra <vohra.sam@gmail.com>

- pull out keepalive from tsa client
- add test coverage for new behaviour

Signed-off-by: Sameer Vohra <vohra.sam@gmail.com>

- Changed the timeout from 5s to 5m as 5s might be too aggresive for
  remote workers that don't have reliable connections.

  This timeout is to detect the condition where we observed SentRequest
  to hang indefinitely, which can still be reliably detected with a
  timeout of 5m.

Signed-off-by: Sameer Vohra <vohra.sam@gmail.com>

Signed-off-by: Sameer Vohra <vohra.sam@gmail.com>

NewRelicConfig has a default value for URL and is never empty

Signed-off-by: Syamala Umamaheswaran <shykart2203@gmail.com>

add configuration for NewRelic insights url

despite us specifying how one would go about configuring the environment
variables for the k8s-topgun tests, we never mentioned how one can go
about getting the kubernetes cluster configured.

`kind` is great as it assumes very few - all you need is docker - and
extending the default configuration is also very great - just supply a
piece of yaml (which is well documented in their website).

Signed-off-by: Ciro S. Costa <cscosta@pivotal.io>
Co-authored-by: Taylor Silva <tsilva@pivotal.io>

k8s-topgun tests are much better run concurrently rather than serially
given how slow the initialization times are compared to unit tests.

Signed-off-by: Ciro S. Costa <cscosta@pivotal.io>

Close sshClient Conn on SendRequest error

Noticed this error message myself and it wasn't clear to me what was
wrong. Someone on discord also mentioned that the error message was not
helpful with their troubleshooting.

This error message should guide the user to look at the likely source of
the error: the parent resource type's config.

Signed-off-by: Taylor Silva <tsilva@pivotal.io>

Co-authored-by: Rui Yang <ryang@pivotal.io>
Signed-off-by: Taylor Silva <tsilva@pivotal.io>

Co-authored-by: Rui Yang <ryang@pivotal.io>
Signed-off-by: Taylor Silva <tsilva@pivotal.io>

Co-authored-by: Rui Yang <ryang@pivotal.io>
Signed-off-by: Taylor Silva <tsilva@pivotal.io>

when --log-db-queries is set to true

Signed-off-by: Rui Yang <ryang@pivotal.io>

contributing: add tiller install instructions

k8s-topgun has a lot of build output whenever any tests fail. Most of
the output is coming from our repeated calls to `kubectl get pods` as we
wait for pods to become ready for testing.

This commit removes our usage of kubectl for this one function in favour
of directly using k8s client-go package. Now this insane amount of json
output will not be present.

Leaving other usages of kubectl alone for now like `kubectl port-forward`.

Signed-off-by: Taylor Silva <tsilva@pivotal.io>

Signed-off-by: Taylor Silva <tsilva@pivotal.io>

the prometheus chart recently updated itself to have
`kube-state-metrics` being a subchart, rather than embedded on itself.

because of that, we need to first fetch its dependencies, and then later
on, install it.

here I also got rid of the Concourse worker that was being deployed -
this is possible because nowadays we offer the ability to do so
(something you couldn't before), and because the worker is unecessary
for this case.

Signed-off-by: Ciro S. Costa <cscosta@pivotal.io>

atc/db: less cryptic error message

Signed-off-by: Taylor Silva <tsilva@pivotal.io>

We should switch this over to using the API directly as well as it can
make the build logs very noisy as well.

Signed-off-by: Taylor Silva <tsilva@pivotal.io>

When using -node, everything in SynchronizedBeforeSuite is only run on
one node, therefore other nodes don't have a pointer to kubeClient, it
was never set on those nodes!

Putting it in the suite's BeforeEach() ensures each node initializes the
kubeClient.

Signed-off-by: Taylor Silva <tsilva@pivotal.io>

Replaced with directly calling the API for the logs of each pod

Signed-off-by: Taylor Silva <tsilva@pivotal.io>

Make k8s-topgun produce less output

Signed-off-by: James Thomson <jthomson@pivotal.io>

k8s/topgun: update prometheus dependencies and disable worker

Using the limit-active-tasks placement strategy gives the perfect opportunity
to dynamically scale workers based on tasks running. This metric makes
it easier for operators to implement such a scaling.

#5057

Co-authored-by: Torben Neufeldt <torben.neufeldt@oss.volkswagen.com>
Signed-off-by: Hannes Hasselbring <hannes@oss.volkswagen.com>

Co-authored-by: Torben Neufeldt <torben.neufeldt@oss.volkswagen.com>
Co-authored-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>
Signed-off-by: Hannes Hasselbring <hannes@oss.volkswagen.com>

Co-authored-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>
Signed-off-by: Hannes Hasselbring <hannes@oss.volkswagen.com>

Co-authored-by: Hannes Hasselbring <hannes@oss.volkswagen.com>
Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>

- To avoid test pollution all previously registered collectors are unregistered in
  garbage collection test

Co-authored-by: Hannes Hasselbring <hannes@oss.volkswagen.com>
Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>

…to be available

- Make metrics test more stable

Co-authored-by: Hannes Hasselbring <hannes@oss.volkswagen.com>
Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>

- introduce wait group so we know the context is cancelled before running the tests

Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>

- Externalize worker polling and status configuration
  to have more stable tests

Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>

Signed-off-by: Syamala Umamaheswaran <shyam@oss.volkswagen.com>