On 03.18.2025 10:50, Zane Healy via groups.io wrote:
My preference for running something like Hercules or SIMH is a Virtual Machine running on an older x86 box with Linux installed. Log in, start up ’screen’ and then start up the emulator. I’ve used a Raspberry Pi for Multics and MVS in the past, but find Virtual Machines to be more convenient. I prefer to run OpenSUSE on the VM, but have also used Ubuntu and Red Hat.
100% agree. I always use Debian for servers and have switched to tmux
instead of screen, but yes, this is how I run all of my Hercules and
simh instances -- Debian VM on my VM server with tmux sessions for each
Hercules (MVS 3.8 / VM/370 / etc) listening on a different console port.
Question, has anyone looked at using Proxmox for this?? I’m looking to migrate my home lab off of ESXI, due to the cost.
Yes. Early this year I made the switch to Proxmox on my VM servers after
a decade of ESXi. Very happy with it.
ONE IMPORTANT TIP, though: I always set the VM CPU to "host" instead of
the default (x86-64-v2-AES) when creating new VMs. I don't have a
cluster of Proxmox servers where I need to worry about live-migrating
VMs between hosts with different CPU models, so it always seemed like a
good idea to me to just pass through the real CPU to VMs instead of
having QEMU fake a specific CPU.
It turns out the *one* VM I accidentally forgot to set the CPU to "host"
when I created it was my mainframe VM that runs all of my Hercules
instances. I was having an intermittent problem where my MVS 3.8
instance wouldn't IPL -- it just hung forever shortly after the IPL
started. No Hercules crash, no errors...just stopped making forward
progress. After a lot of troubleshooting, I eventually realized the VM
CPU was set to x86-64-v2-AES, not "host". I changed the CPU model of
the VM to "host". Sure enough, the intermittent IPL problem went away!
Hercules has been reliable again after I made that change.
-Matthew
