Update and expand using kubernetes quick doc

buckaroogeek commented

2024-07-30 20:51:55 +00:00

Contributor

Significant revision of using kubernetes quick doc. Partitioned doc into top level page a 6 nested pages (update to navigation too). Added content.

New versioned kubernetes rpms (along with cri-o and cri-tools rpms) are coming to rawhide. Kubernetes, for example will now have 4 versions in rawhide instead of just 1 version. The legacy plain kubernetes will still exist in rawhide but will not exist in F42.

These changes are explained in this quick doc revision.

The first question is: does this new content still meet quick doc standards and purpose?

Significant revision of using kubernetes quick doc. Partitioned doc into top level page a 6 nested pages (update to navigation too). Added content. New versioned kubernetes rpms (along with cri-o and cri-tools rpms) are coming to rawhide. Kubernetes, for example will now have 4 versions in rawhide instead of just 1 version. The legacy plain kubernetes will still exist in rawhide but will not exist in F42. These changes are explained in this quick doc revision. The first question is: does this new content still meet quick doc standards and purpose?

jasonbrooks commented

2024-08-09 17:07:01 +00:00

First-time contributor

I tested the "Creating a Kubernetes cluster on Fedora" doc. I ran through the steps on a f40 cloud edition VM. They worked pretty well, with a few exceptions that might be due to the configuration of the cloud edition.

swap-create@zram0.service and firewalld.service weren't enabled on the cloud edition so I got (ignorable) errors on these, this may or may not be worth noting.

All my pods came up, with the exception of the coredns pods, which kept crashlooping, with an error citing this page. The issue has to do with systemd-resolved, and I found a fix here.

sudo mkdir -p /etc/systemd/resolved.conf.d/
sudo cat <<EOF | sudo tee /etc/systemd/resolved.conf.d/stub-listener.conf
[Resolve]
DNSStubListener=no
EOF

I tested the "Creating a Kubernetes cluster on Fedora" doc. I ran through the steps on a f40 cloud edition VM. They worked pretty well, with a few exceptions that might be due to the configuration of the cloud edition. `swap-create@zram0.service` and `firewalld.service` weren't enabled on the cloud edition so I got (ignorable) errors on these, this may or may not be worth noting. All my pods came up, with the exception of the coredns pods, which kept crashlooping, with an error citing [this page](https://coredns.io/plugins/loop/#troubleshooting). The issue has to do with systemd-resolved, and I found a fix [here](https://fedoraproject.org/wiki/Changes/systemd-resolved#Local_stub_resolver_on_port_53). ``` sudo mkdir -p /etc/systemd/resolved.conf.d/ sudo cat <<EOF | sudo tee /etc/systemd/resolved.conf.d/stub-listener.conf [Resolve] DNSStubListener=no EOF ```

buckaroogeek commented

2024-08-09 21:19:33 +00:00

Author

Contributor

I tested the "Creating a Kubernetes cluster on Fedora" doc. I ran through the steps on a f40 cloud edition VM. They worked pretty well, with a few exceptions that might be due to the configuration of the cloud edition.

Wonderful. Thank you very much.

I will update notes for swap and firewall that these services may not be running. I typically test in vagrant hosted vms using the vagrant box that the cloud team prepares so see this also.

Thanks for the fix the coredns crash loop. I will give it a try. I have also found that the when I use a kubelet config file instead of the systemd based command line configuration (systemd is the default in the non-versioned rpms) that I do not get a crash loop. In any case, I really appreciate the work to find a solution. I will amend the document to include it as an option. Another solution that I used is to amend the coredns config file using kubectl and changing the dns entry to a known dns IP for the lan the cluster is running on.

> I tested the "Creating a Kubernetes cluster on Fedora" doc. I ran through the steps on a f40 cloud edition VM. They worked pretty well, with a few exceptions that might be due to the configuration of the cloud edition. > Wonderful. Thank you very much. I will update notes for swap and firewall that these services may not be running. I typically test in vagrant hosted vms using the vagrant box that the cloud team prepares so see this also. Thanks for the fix the coredns crash loop. I will give it a try. I have also found that the when I use a kubelet config file instead of the systemd based command line configuration (systemd is the default in the non-versioned rpms) that I do not get a crash loop. In any case, I really appreciate the work to find a solution. I will amend the document to include it as an option. Another solution that I used is to amend the coredns config file using kubectl and changing the dns entry to a known dns IP for the lan the cluster is running on.

buckaroogeek commented

2024-08-19 18:20:11 +00:00

Author

Contributor

1 new commit added