I've wanted to start working with Test Containers for a long time. This is really helpful to solve some questions I had about using it. Much appreciated.
How do you handle port conflicts? If you are running the containers from docker compose when developing so that they are running when you run the test. Wont you get an error saying that the port is already allocated?
That's not my type of workflow. I test-drive first and then I might run docker-compose manually for manual testing. So, I never got to a port conflict. However, in your use case, I would create an override like is explained here: mindbyte.nl/2018/04/04/overwrite-ports-in-docker-compose.html
Hello :) thanks for this video that provides practical example. I'm also interested in the implementation of integration tests with SQL Server Database during CI. Once my Azure pipeline is running I'm facing issues. Seems my connection string doesn't find the SQL Server :/ Any idea? Kind regards
Hi! Are you running the SqlServer in a container? Are you using the docker-compose as I show in the video? Can you see in the pipeline logs that the SQL Server started successfully?
Hi Gui. I downloaded the code for this and it works really well. Thanks for sharing. 🙂 I wanted to ask if in your experience you shared the same docker container across all the integration tests or you create a container for each one of the test classes.
@@gui.ferreira - I created for my use case a global strategy for NUnit. I have a global setup and tear down for the assembly. The two methods take care to create the container before running the tests and destroy it after running the tests. Do you have any raccommendation on how to speed them both locally and on Azure?
@@DJEAUK I'm not experienced in NUnit but looks like you are doing exactly what I'm saying. The only difference may be that you are relating that to the assembly, and with xUnit, you may have multiple Fixtures on your tests. Regarding speed optimizations, the only recommendation is to make sure you pin container versions and don't bring stuff into the docker-compose file that is not required. Besides that, I know that some platforms have a caching system for container images. I think that I've seen that in GitHub Actions.
@@gui.ferreira - thanks for you feedback. Using the overriding capability of docker compose, I keep multiple docker compose files and orchestrate the stack composition with a bash script. Therefore, I have docker compose file per resource. Using the approach you shared in the video, now I can load a single docker compose file for the integration tests, in the specific the database server. For CI/CD, I have a requirement of using Azure DevOps Pipelines. I need to see if they offer a caching system in the hosted agents. Thanks for sharing your idea. It helped me a lot. Keep up with the good work!
how do you handle kafka (or rabbitmq) server startup delay in `docker-compose.yaml` or at fixature level? I often end up running and crashing web server first even after using `depend_on` attribute in `docker-compose.yaml`
That is one of the Advantages of using Testcontainers directly since it has wait strategies in place. For the cases you mentioned, if you are using FluentDocker, you can Wait for an HTTP response. Both Kafka and Rabbit can expose an HTTP Port
![ทุกครั้งที่หลับตา (Lucid Dream) - AYLA's [ Official MV ]](http://i.ytimg.com/vi/4i1OBAQyv58/mqdefault.jpg)
Today I learned: test containers. Thanks for the lead to the concept and the modifications!
Glad it was helpful!
I've wanted to start working with Test Containers for a long time. This is really helpful to solve some questions I had about using it. Much appreciated.
Glad to hear. Keep in mind that what I show here is an alternative to the common practice when using TestContainers.
@@gui.ferreira Did you ever try running test containters in a build pipeline like GitHub actions?
@@MilanJovanovicTech Not on GitHub Actions, but on TeamCity.
Top content, I'm going to use it in my projects. Thank you!
Thanks, Vadym! 🙏
Let me know how did it go!
How do you handle port conflicts? If you are running the containers from docker compose when developing so that they are running when you run the test. Wont you get an error saying that the port is already allocated?
That's not my type of workflow. I test-drive first and then I might run docker-compose manually for manual testing. So, I never got to a port conflict.
However, in your use case, I would create an override like is explained here: mindbyte.nl/2018/04/04/overwrite-ports-in-docker-compose.html
Hello :) thanks for this video that provides practical example. I'm also interested in the implementation of integration tests with SQL Server Database during CI. Once my Azure pipeline is running I'm facing issues. Seems my connection string doesn't find the SQL Server :/ Any idea? Kind regards
Hi! Are you running the SqlServer in a container? Are you using the docker-compose as I show in the video? Can you see in the pipeline logs that the SQL Server started successfully?
Hi Gui. I downloaded the code for this and it works really well. Thanks for sharing. 🙂 I wanted to ask if in your experience you shared the same docker container across all the integration tests or you create a container for each one of the test classes.
Hi! Thanks 🙏
I use the same Test Fixture on all the test classes that will need those containers, so they run faster.
@@gui.ferreira - I created for my use case a global strategy for NUnit. I have a global setup and tear down for the assembly. The two methods take care to create the container before running the tests and destroy it after running the tests. Do you have any raccommendation on how to speed them both locally and on Azure?
@@DJEAUK I'm not experienced in NUnit but looks like you are doing exactly what I'm saying. The only difference may be that you are relating that to the assembly, and with xUnit, you may have multiple Fixtures on your tests.
Regarding speed optimizations, the only recommendation is to make sure you pin container versions and don't bring stuff into the docker-compose file that is not required. Besides that, I know that some platforms have a caching system for container images. I think that I've seen that in GitHub Actions.
@@gui.ferreira - thanks for you feedback. Using the overriding capability of docker compose, I keep multiple docker compose files and orchestrate the stack composition with a bash script. Therefore, I have docker compose file per resource. Using the approach you shared in the video, now I can load a single docker compose file for the integration tests, in the specific the database server. For CI/CD, I have a requirement of using Azure DevOps Pipelines. I need to see if they offer a caching system in the hosted agents.
Thanks for sharing your idea. It helped me a lot.
Keep up with the good work!
@@DJEAUK Thanks for the support 🙏
how do you handle kafka (or rabbitmq) server startup delay in `docker-compose.yaml` or at fixature level? I often end up running and crashing web server first even after using `depend_on` attribute in `docker-compose.yaml`
That is one of the Advantages of using Testcontainers directly since it has wait strategies in place.
For the cases you mentioned, if you are using FluentDocker, you can Wait for an HTTP response. Both Kafka and Rabbit can expose an HTTP Port