Sometimes with virtual machines, it’s necessary to have a watchdog timer that monitors the VM and automatically restarts it in the case of any serious problems such as hangs. I have that need, and need such a configuration for both KVM and for Proxmox guests. Fortunately this is easy to set up.

Leaning on vmWare PowerCLI for much of my day-to-day system administration needs, I also have need for running PowerCLI scripts regularly. There are many ways to do this, but since I have an OpenShift cluster available, why not run those scripts in containers so you can simply set up a cronjob manifest to run them?

A quick snippet. Sometimes it is necessary to have a very old browser kicking around somewhere with all the security hooks turned off and with an old version of Java on it. The usual use case for this is getting access to some kind of embedded console, and you certainly do not want to be using that as your standard browser. A throwaway container that can run the browser for you on demand is exactly what is needed in that situation.

What a mess. I’d upgraded my work laptop from Windows 10 to Windows 11, and discovered some strangeness with OpenSSH agent forwarding when using Windows’ built-in ssh client. Mostly, I use ssh through WSL2, but VS Code obviously runs on the Windows layer, meaning that agent forwarding has to Just Work. Which it did not. I kept getting this sort of nonsense when trying to use a forwarded agent;

Turns out it’s been nearly a year since I’ve posted. Whoops.

MetalLB is a load balancer for Kubernetes that allows direct external IP ingress into Kubernetes clusters for access to pods and services. Whereas a product like Multus adds interfaces to pods directly (thus giving them layer 2 access to that network), MetalLB implements the Kubernetes LoadBalancer kind, permitting services to be accessed directly from outside the cluster on their own IP addresses.