Of course, the recent annoucement of the ClusterHAT on the Raspberry Pi website caught my attention. I grew up with (VAX-)clusters, so I definitely needed to explore this. I already operate a VAXcluster on a collection of Raspberries, described here and in previous blogs.
The nifty device interfaces a (controller) Raspberry Pi A+/B+/2/3 with up to 4 Raspberry Pi Zeros in a compact form factor.
As it turns out, assembling and configuring the HAT is straightforward if your know your Raspberries. It leaves you -networkwise- with the Zeroes bridged to the controller Pi. Physically, it is the black USB cable shown in the video that carries the “inside” network traffic. The bridge on the controller Pi allows access to the external world through the Ethernet port.
Now my simh VAXcluster needs a separate network interface on Linux for each member. So on the controller Pi I added a tap0 interface to the existing bridge to be used by the VAX-11/780 cluster boot member. Each (MicroVAX II) LAVC satellite member is on its own Pi Zero, to I added a bridge on those as well, connecting the usb0 and tap0 interfaces to it.
I copied the simh disk images from my existing VAXcluster to the new platform, and there it was: A VAXcluster running on a Raspberry Pi with ClusterHAT. One quote on the ClusterHAT website stood out: “Controller Pi can be rebooted without interrupting power to Pi Zeros (network recovers on boot)”. So with everything running, I boldy restarted the controller Pi. After a few minutes, the MicroVAX II instances started responding again. They had been waiting in cluster state transition for the boot member with the single vote to reappear, and so it did. The cluster as such was still in place. VAXclusters are so resilient!
Thank you Chris Burton for this experience!
Get your own simh stuff from the github repository!