this is a very, very useful article, accurate and easy to follow.
K8s is a fast development project, which leaves many other article hard to follow, and setting up correctly. Glad I found this one.
While Selenium Hub is a single-point-of-failure, the recovery time of a failure is almost instant because Kubernetes will schedule a replacement container right away.
Sometimes removing a single-point-of-failure results in a more confusing and difficult to operate system. I didn't want to take on that risk.
In my situation, I had no need for HA (high availability) because the mean-time-to-recovery of a Selenium Hub is so fast and without any real consequence.
I think you should review your requirements to see if you really need a HA Hub.
For all the chrome nodes the VNC port is 5555 and its connected to localIP. In this case how do we access the console. For example if we install this selenium grid on an ec2 machine then how we access the VNC desktop via public IP?
Comments
Very well written article. Thanks for exposing so many good links resources, I honestly struggled to find any around.
export
You are very welcome! I hope this helps. Feel free to reach out over email if you have any specific questions.
export
Has someone tried HPA for selenium node. I not able to scale up as the CPU consumption is very low.
export
What does HPA mean in the context of Selenium node?
export
I did try and you can see my talk here - https://www.slideshare.net/aragavan/selenium-grid-based-on-kubernetes-docker-traefik-and-helm (slides 19-21) but not yet 100% successful. My issue is, Kubernetes still not able to get the mertics even though metrics server is running fine. @russel
export
this is a very, very useful article, accurate and easy to follow. K8s is a fast development project, which leaves many other article hard to follow, and setting up correctly. Glad I found this one.
export
thanks for such a detailed explanation. this is very useful information. Saved a lot of time and effort researching.
export
Hi Nice Article. But How do I scale Selenium HUB because that is single point of failure.
export
While Selenium Hub is a single-point-of-failure, the recovery time of a failure is almost instant because Kubernetes will schedule a replacement container right away.
Sometimes removing a single-point-of-failure results in a more confusing and difficult to operate system. I didn't want to take on that risk.
In my situation, I had no need for HA (high availability) because the mean-time-to-recovery of a Selenium Hub is so fast and without any real consequence.
I think you should review your requirements to see if you really need a HA Hub.
export
For all the chrome nodes the VNC port is 5555 and its connected to localIP. In this case how do we access the console. For example if we install this selenium grid on an ec2 machine then how we access the VNC desktop via public IP?
export
Is it possible to attach physical Mac to a Selenium grid on kubernetes to run appium test on iOS simulators ? thank you
export
It is very simple and informative
export