Amazing work ! thank you for all the awesome videos. I have a small request, can you do a video of a Platform Agnostic Installation please ? without UPI or IPI. Thanks again for all the hard work you're making.
Thanks for this video. I'm following this video but I have issue. When start Bootstrap node or any node of Openshift Cluster, fail with this problem: emergency.service failed with result "exit-code". How to fix that problem? Thanks again!
Usually this is a network issue, either the load balancer isn't configured correctly or DNS isn't resolving the api. Double check both of these and add "--log-level=debug" to your install script.
Hello OCPdude, Amazing video. If I want to use two more nodes as infra, I upload it with the worker's ingnition and after that how would I create its machine config pool? Is it necessary to create a machine config for infra?
Hi OCPdude, Amazing video. But will be good if you have detail steps too on text & picture, like a your github or website. So we can deep learning your steps too. Special note when you add configuration parameter when create vm's, the text it's not clear. Please inform me if already exist. I'm very appreciate for this. 🙏 Thank you.
You an adjust the video up to 4K, hopefully that will help the clarity. I did not write out the steps from this video, but you can refer to my OKD install github.com/ocpdude/okd-install since it has many of the same steps.
thanks for your video! quick question, is vSAN required? If my VM cluster only has 3 physical machines without vSAN/ separated NFS server, is it still possible to install OCP on VMWare using VMware vSphere, not Bare metal to provision the instances? Thanks!
Yes, vSan is not a requirement. I actually only use it for the workers and shared storage. Previously I just pinned the nodes on hosts and leveraged NFS.
Hi OCPdude. Very nice turotials, thanks! The bootstrap VM in my setup always goes into Emergency Shell then hang after 'Started Setup Virtual Console'. Any pointer to what's wrong? My install-config.yaml looks similar to yours with the proxy block since my lab is behind proxy.
@@OCPdude actually we are not able to touch our production DNS nor we can create DNS and DHCP on same up subnets provided by VMware team . I want to know in detail how can in use DNS and DHCP in restricted network, where external users will access cluster services
@@ajaydurande8389 If you use static IP's like in this video, you won't need DHCP. As for DNS, at the very least you'll need some way of resolving the api and *.apps. As to not conflict with other domains, you could use a sub-domain for DNS resolution, or you could just use host files from your build and access machine (bastion host).
@@OCPdude bootstrap and the 3 of master are normally ignition success, but the internal server error continues to occur due to the worker node I checked ignition success of all masters and turned on the worker right away
@@OCPdude I don't think there is a problem with api or dns, so wouldn't the master's ignition also fail? Of course, I checked dns resolve with tools such as dig and also checked the port that is waiting(tcp) on the api server(haproxy)
@@hyunjunechoi1154 bootstrap kicks off the masters, the workers load from the master nodes (api) and require access to port 6443 over api.cluster.domain.com, api-int.cluster.domain.com and port 443 for *.apps.cluster.domain.com. ESX should also be version 6.7U3 or 7, hardware level 15. Some things to check.
I'm sorry for that naive question, but how do you configure the haproxy box with two IP addresses? I think you can't put two IPs from the same subnet to one Linux machine, am I right?
@@emadkamel2801 I really only have 2 reasons... 1) I can vMotion the compute nodes as they don't have the same latency demands that the controller nodes do and 2) I have a lot of capacity there so scaling up nodes is easier.
@@OCPdude I thought it was related to the bug you reported to red hat regarding provisioning PVC (VMware vSphere 7 cannot create "thin" pvc's after a clean install.), so you moved worker nodes to vSAN to be able to provision PVCs successfully. I'm stuck now on the same case, when I create a PVC it is stuck on pending state and gives me the same error you faced (Failed to provision volume with StorageClass "thin": ServerFaultCode: Cannot complete file creation operation.)
Hey OCPdude do you know if is possible to migrate the HAproxy for the api and ingress of a UPI install to a deployment inside the OCP installation as the one that the IPI install provides ?
Roberto, Unfortunately I wasn't able to quickly locate the internal proxy configuration used in IPI though, it does appear to be VIP's in haproxy. I discovered this by /etc/keepalived/keepalived.conf on the master nodes and /etc/kubernetes/static-pod-resources/keepalived/* & haproxy/ - sorry I don't have a good configuration answer for you at this point.
@@OCPdude Dont worry thats ok, i can handle the ha proxy config, i like UPI it gives me more flexibility but when the numbre of apps start to increase the haproxy config for the required nodeports start to grow and is hard to maintain
I'd recommend you look at Code Ready Containers (CRC) instead : ruclips.net/video/xoQvs8HD5xc/видео.html You'll get the same environment without taxing your system so much, and all on 1 node.
@@OCPdude Thanks for the suggestion. You won't believe but I actually bought a laptop with 64gb ram for just testing the OC installation :) . And I tried installing (almost all versions from 4.5 to 4.10) it more than 50 times using Linux hypervisor VMM and it never worked. Finally I came across your video with different way of installation so thought of checking with you before I try.
@@neerajsharma3690 right, nice! I haven’t done it myself using workstation. I figure networking might be trouble, and you will still need a load balancer and good local dns. You can also run OCP as a single node, unless your goal is a full cluster. With enough CPU you should be able to pull off a cluster and 2 worker nodes with the memory you have. Although your CPU will likely hate you.
Hi OCPdude, Provided guestinfo.afterburn.initrd.network-kargs parameter with required values IP=**** but after booting the bootstarp node somehow its taking ip v6 and local domain instead of provided one. Any help on this?
Thank you for this video - exactly what I was looking for! Subscribed
Good job !! Great video, Thanks.
Thank you, the best guide!
Amazing work ! thank you for all the awesome videos.
I have a small request, can you do a video of a Platform Agnostic Installation please ? without UPI or IPI.
Thanks again for all the hard work you're making.
Great video 👍
Thanks for this video.
I'm following this video but I have issue. When start Bootstrap node or any node of Openshift Cluster, fail with this problem: emergency.service failed with result "exit-code".
How to fix that problem?
Thanks again!
Usually this is a network issue, either the load balancer isn't configured correctly or DNS isn't resolving the api. Double check both of these and add "--log-level=debug" to your install script.
Hello OCPdude, Amazing video. If I want to use two more nodes as infra, I upload it with the worker's ingnition and after that how would I create its machine config pool? Is it necessary to create a machine config for infra?
Hi OCPdude, Amazing video.
But will be good if you have detail steps too on text & picture, like a your github or website. So we can deep learning your steps too. Special note when you add configuration parameter when create vm's, the text it's not clear.
Please inform me if already exist. I'm very appreciate for this. 🙏
Thank you.
You an adjust the video up to 4K, hopefully that will help the clarity. I did not write out the steps from this video, but you can refer to my OKD install github.com/ocpdude/okd-install since it has many of the same steps.
thanks for your video! quick question, is vSAN required? If my VM cluster only has 3 physical machines without vSAN/ separated NFS server, is it still possible to install OCP on VMWare using VMware vSphere, not Bare metal to provision the instances? Thanks!
Yes, vSan is not a requirement. I actually only use it for the workers and shared storage. Previously I just pinned the nodes on hosts and leveraged NFS.
Is there a reason why don't you use "openshift-install waiting-for bootstrap-complete" unlike the manual?
I’ve installed it so often I assume it will complete. 😬
Hi OCPdude. Very nice turotials, thanks! The bootstrap VM in my setup always goes into Emergency Shell then hang after 'Started Setup Virtual Console'. Any pointer to what's wrong? My install-config.yaml looks similar to yours with the proxy block since my lab is behind proxy.
Is it loading the append-bootstrap.yaml?
@@OCPdude I don't think it's getting to bootstrapping part yet. Here is the screenshot.
sorry, I couldn't post the screenshot.
@@vanpham2527 email me.
Can you please make a video on openshift install in restricted network on vsphere using UPI Method
I assume you'd just combine the disconnected registry demo along with the vmware upi install video. Is there something else you're looking for?
@@OCPdude actually we are not able to touch our production DNS nor we can create DNS and DHCP on same up subnets provided by VMware team . I want to know in detail how can in use DNS and DHCP in restricted network, where external users will access cluster services
@@ajaydurande8389 If you use static IP's like in this video, you won't need DHCP. As for DNS, at the very least you'll need some way of resolving the api and *.apps. As to not conflict with other domains, you could use a sub-domain for DNS resolution, or you could just use host files from your build and access machine (bastion host).
can i specifiy the ips for the master node in yaml file as i am not using dhcp
@@_l0v3saki you can in the install-config file.
I'm curious about the latest version that was successfully installed on vshpere
I try 4.10 but installation failed
I run 4.10 OKD today without any issues.
@@OCPdude bootstrap and the 3 of master are normally ignition success, but the internal server error continues to occur due to the worker node
I checked ignition success of all masters and turned on the worker right away
@@hyunjunechoi1154 check your load balancer and DNS, your api and wildcard should resolve correctly.
@@OCPdude I don't think there is a problem with api or dns, so wouldn't the master's ignition also fail?
Of course, I checked dns resolve with tools such as dig and also checked the port that is waiting(tcp) on the api server(haproxy)
@@hyunjunechoi1154 bootstrap kicks off the masters, the workers load from the master nodes (api) and require access to port 6443 over api.cluster.domain.com, api-int.cluster.domain.com and port 443 for *.apps.cluster.domain.com. ESX should also be version 6.7U3 or 7, hardware level 15. Some things to check.
I'm sorry for that naive question, but how do you configure the haproxy box with two IP addresses? I think you can't put two IPs from the same subnet to one Linux machine, am I right?
Sub interface. You can Linux or Windows listening on multiple IP address, even on the same subnet.
@@OCPdude Thanks shaker ❤❤
Hello Shaker, I've a question. what is the point from provisioning worker nodes on vSAN?
@@emadkamel2801 I really only have 2 reasons... 1) I can vMotion the compute nodes as they don't have the same latency demands that the controller nodes do and 2) I have a lot of capacity there so scaling up nodes is easier.
@@OCPdude I thought it was related to the bug you reported to red hat regarding provisioning PVC (VMware vSphere 7 cannot create "thin" pvc's after a clean install.), so you moved worker nodes to vSAN to be able to provision PVCs successfully.
I'm stuck now on the same case, when I create a PVC it is stuck on pending state and gives me the same error you faced (Failed to provision volume with StorageClass "thin": ServerFaultCode: Cannot complete file creation operation.)
Hey OCPdude do you know if is possible to migrate the HAproxy for the api and ingress of a UPI install to a deployment inside the OCP installation as the one that the IPI install provides ?
Yeah, it shouldn’t be a problem to just run it in the platform. I’m doing a rebuild tomorrow, I’ll give it a go and get back to you.
Roberto, Unfortunately I wasn't able to quickly locate the internal proxy configuration used in IPI though, it does appear to be VIP's in haproxy. I discovered this by /etc/keepalived/keepalived.conf on the master nodes and /etc/kubernetes/static-pod-resources/keepalived/* & haproxy/ - sorry I don't have a good configuration answer for you at this point.
@@OCPdude Dont worry thats ok, i can handle the ha proxy config, i like UPI it gives me more flexibility but when the numbre of apps start to increase the haproxy config for the required nodeports start to grow and is hard to maintain
@@bobby07241
youtube.com/@OCPdude
Hi guys, did you find how to move Haproxy inside the cluster rather it being on provisioner node?
@@OCPdude Hi, did you find any solution to move *api *apps from Haproxy to inside the cluster itself?
Can I setup this lab in VMware workstation 16?
I'd recommend you look at Code Ready Containers (CRC) instead : ruclips.net/video/xoQvs8HD5xc/видео.html You'll get the same environment without taxing your system so much, and all on 1 node.
@@OCPdude Thanks for the suggestion. You won't believe but I actually bought a laptop with 64gb ram for just testing the OC installation :) . And I tried installing (almost all versions from 4.5 to 4.10) it more than 50 times using Linux hypervisor VMM and it never worked. Finally I came across your video with different way of installation so thought of checking with you before I try.
@@neerajsharma3690 right, nice! I haven’t done it myself using workstation. I figure networking might be trouble, and you will still need a load balancer and good local dns. You can also run OCP as a single node, unless your goal is a full cluster. With enough CPU you should be able to pull off a cluster and 2 worker nodes with the memory you have. Although your CPU will likely hate you.
What is the role of Bastion host?
👋 the bastion host is just where I am installing OpenShift from, it is not relevant for the install.
@@OCPdude how you able to get cluster information from bastion host while it was not the part of cluster?
@@ridinglycan318 by exporting the .kube/config 21:08 which has the relevant info for the cluster.
Font is not clear
its clear, make it HD while viewing.
Bro still not able to view
RUclips is still processing this video into HD and then 4K… I have unlisted it until they are done.
Hi OCPdude, Provided guestinfo.afterburn.initrd.network-kargs parameter with required values IP=**** but after booting the bootstarp node somehow its taking ip v6 and local domain instead of provided one. Any help on this?
Same issue with me are any one know how to solve this @ocpdude
@@mohamedabdallah6493 same issue for me , any suggestion