It would be much easier to follow along if a simple architecture diagram is provided showing the various IP addresses used in the lab, which IP address is belonging to what server. Right now, is very difficult to visualize what is happening.
Great content, I worked with Kubernetes a couple of years ago and was stuck by the "start-from-zero" difficulties. Glad to see things have improved a lot. Couple of questions for you: 1) The company I worked with 2 years ago was a Rancher user; I didn't really care much for Rancher (granted they had an older version of Rancher), do you think it's still useful? 2) What's your opinion on the vCluster project? One of the major issues with Kube I encountered was weak namespace isolation.
Solomon, I love Rancher and use it regularly...good stuff. Also, i need to check into the vCluster project. I think I have played around with it in the past but definitely need to revisit this one. Thanks again for your comment!
Nice video!!! I will test this out. Can you do a follow up to complete the setup to add worker nodes? Then add management WebUI interface like Rancher or Portainer and finally how can roll out any docker apps in the cluster?. Also can the worker nodes can be on a different subnet / VLAN from the control plane nodes?
You could convert docker-compose stacks to Kubernetes manifests, but that is not really advised if you you're a novice. Start with using Lens, Rancher is pretty intrusive in your cluster and can't easily be removed from it. Start learning the deployment syntax, as well as what PVCs and ingresses are and go the cli + manifests route. If you just use the GUIs you'll never really understand what's going on and what the differences between deployment, statefulsets, replicasets and pods are. It's hard work! You could VLAN off your workers, but that's pretty much useless as you need full access between all the nodes (Controlplane + workers) for the overlay network, healthchecks etc to work.
Alex, thank you for your comment! Try them all if you have a chance...that's the great thing about learning you can try many different things and see what works best for you...keep us posted on hiw it turns out 👍
really good content ⭐⭐⭐⭐ I've learned few news tools \techs form this channel already A place for improve is to do something with chapters: divided to often, as for me, and the names are not the clearest description of what you looking for, if you looking for table-of-contents to return to something
When I tear down my cluster, one of the VMs retains the aliased IP address on the ens192 interface. Is there something I should be doing that will automagically clean that up as well?
I need help. I have K8s cluster with four k8s nodes. Would I be able to use K3sup with the k8s cluster nodes? If not how would I setup the three needed K3s nodes? I want to spin up new nodes (either k3s or k8s) because the other four originally were setup by someone else and I dont want to mess with the configurations if I dont need to
@adriandeveraaa Let's discuss more on the VHT forums. Sign up and we can create a topic to discuss further there: www.virtualizationhowto.com/community
@@VirtualizationHowto On your blog, the dashes in the commands copy as one instead of two. Took me a bit to figure out what was wrong as the alias command would take but I kept getting errors when trying to start the manifest.
What is the ens192 interface in the kube-vip manifest? The kube-vip pod won't start unless I remove that from the manifest: level=fatal msg="ens192 is not valid interface, reason: get ens192 failed, error: Link not found"
@@VirtualizationHowto ah, got it. I wasn't sure if it was binding it to an interface, or naming some virtual interface created and used by it. Thanks for the clarification.
It would be much easier to follow along if a simple architecture diagram is provided showing the various IP addresses used in the lab, which IP address is belonging to what server. Right now, is very difficult to visualize what is happening.
Great content, I worked with Kubernetes a couple of years ago and was stuck by the "start-from-zero" difficulties. Glad to see things have improved a lot. Couple of questions for you:
1) The company I worked with 2 years ago was a Rancher user; I didn't really care much for Rancher (granted they had an older version of Rancher), do you think it's still useful?
2) What's your opinion on the vCluster project? One of the major issues with Kube I encountered was weak namespace isolation.
Solomon, I love Rancher and use it regularly...good stuff. Also, i need to check into the vCluster project. I think I have played around with it in the past but definitely need to revisit this one. Thanks again for your comment!
Thanks for the great video! If I start adding worker nodes, do I also have to install kube-vip on those too?
Nice video!!! I will test this out. Can you do a follow up to complete the setup to add worker nodes? Then add management WebUI interface like Rancher or Portainer and finally how can roll out any docker apps in the cluster?. Also can the worker nodes can be on a different subnet / VLAN from the control plane nodes?
You could convert docker-compose stacks to Kubernetes manifests, but that is not really advised if you you're a novice. Start with using Lens, Rancher is pretty intrusive in your cluster and can't easily be removed from it. Start learning the deployment syntax, as well as what PVCs and ingresses are and go the cli + manifests route. If you just use the GUIs you'll never really understand what's going on and what the differences between deployment, statefulsets, replicasets and pods are. It's hard work!
You could VLAN off your workers, but that's pretty much useless as you need full access between all the nodes (Controlplane + workers) for the overlay network, healthchecks etc to work.
Good stuff. I'm stuck trying to decide which youtuber's HA k3s solution to use, though. :)
Alex, thank you for your comment! Try them all if you have a chance...that's the great thing about learning you can try many different things and see what works best for you...keep us posted on hiw it turns out 👍
really good content ⭐⭐⭐⭐
I've learned few news tools \techs form this channel already
A place for improve is to do something with chapters: divided to often, as for me, and the names are not the clearest description of what you looking for, if you looking for table-of-contents to return to something
why is your air conditioner in the back so close to the desk? Why so low?
keep it up!!
Thank you!
When I tear down my cluster, one of the VMs retains the aliased IP address on the ens192 interface. Is there something I should be doing that will automagically clean that up as well?
I need help. I have K8s cluster with four k8s nodes. Would I be able to use K3sup with the k8s cluster nodes? If not how would I setup the three needed K3s nodes? I want to spin up new nodes (either k3s or k8s) because the other four originally were setup by someone else and I dont want to mess with the configurations if I dont need to
@adriandeveraaa Let's discuss more on the VHT forums. Sign up and we can create a topic to discuss further there: www.virtualizationhowto.com/community
Just made a post. Thanks! @@VirtualizationHowto
When using multiple servers in this configuration, do you configure kube-vip individually or does that get pushed when they join to the master server?
Marc thanks for the comment! Yes that is the case, when you join the cluster with the additional nodes, you use the VIP IP address for the cluster.
@@VirtualizationHowto On your blog, the dashes in the commands copy as one instead of two. Took me a bit to figure out what was wrong as the alias command would take but I kept getting errors when trying to start the manifest.
What is the ens192 interface in the kube-vip manifest? The kube-vip pod won't start unless I remove that from the manifest:
level=fatal msg="ens192 is not valid interface, reason: get ens192 failed, error: Link not found"
Nate, this is an option you can use to specify a specific network adapter. This is the network adapter in my VM. Glad you caught that.
@@VirtualizationHowto ah, got it. I wasn't sure if it was binding it to an interface, or naming some virtual interface created and used by it. Thanks for the clarification.