GRPC Testing with K6

Here we talk about setting up our sandboxed environment, ready to start creating our tests.  This will include a set of gRPC tests.

In this lecture we talk about setting up the gRPC web services, so we can run our K6 tests against them.  I talk about the proto files, the service an the implimentation.

Run a test

Finally we need to run a test to ensure that we are good to start the rest of the course and build our own gRPC tests. 

Here is some instruction for windows only users

In this lecture we will talk about two teams in  very different locations and time zones.  What implications this has for the teams and what we can do to reduce the dependancy upon each other.  Inter-dependancy can stop teams from focusing on their core objectives and can diminish time as developers wait to gain clarity. 

Smoke testing offers clarity through implementation code, confidence through repeatability and self empowerment.

In this lecture we talk about the value that Reliability, Repeatability and Confidence can provide your team and business.

Regression is something that we need to protect against and a good set of automated tests run against as system on every commit can help prevent regression.  In this lecture we discuss what regression is and how smoke test can reduce regression.

Lets make our first smoke test using K6 and start a journey to automated testing

We look at what and how to soak test and what the benefits of this type of tests are.

We look at what and how to load test and what the benefits of this type of tests are.

We look at what and how to spike test and what the benefits of this type of tests are.

We look at how to handle errors that don't seem to be handled by the k6 framework counts.

We look at the response object in its entirety

We look at two types of assertions check and Describe.

We start to create multiple test to include in our framework

Here we create a home for our smoke tests.

Lets look at reusing code for our checks

Here we focus on tests that test our failures and return positive results as expected behaviour

I explore environment vars, protecting sensitive data and authentication.

Here we add success and failure counters

We explore how to handle "google.protobuf.any" type

We now expand the any type to make our code data driven via a json object

Here we explore the OneOf proto Type

We add our test framework to jenkins and run it

We now modify our framework to abort a test run on the first failure and add a no colour mode

We add our framework to a kubernetes job using minikube

We explore staging multiple calls and passing returned data down the chain

We now add these staged calls to our framework and talk about how we can make them stand out from the other tests

Here we explore sharing one connection over multiple users at a concurrent rate of 3000

Here we show how to give each of the 3000 users their own connection, once and we look at he results side by side with the one connection for all tests and one connection per iteration of the tests.

Here we speak about adding random arguments to our gRPC calls on a per iteration bases to mimic what would happen in a live environment.

Here we speak about putting a threshold in and using our error counter from inside and outside of k6 to stop the performance test if they have went bad.

I no further refine the one connection per user code and further expand on our knowledge of init code.