Downloads
Pre-built SocTalk demo VM images, published as GitHub Release artifacts on every v* tag of the soctalk/soctalk repository.
All images are the same Ubuntu 24.04 + K3s + setup wizard build, just converted to different virtual-disk formats. Pick the format your hypervisor consumes natively. There is no per-format functional difference.
Latest release
Current version: v0.2.1: release page · all releases
The release includes:
soctalk-demo-0.2.1.qcow2.xz, KVM, QEMU, libvirt, Proxmoxsoctalk-demo-0.2.1.vmdk.xz, VMware ESXi, Workstation, Fusion, VirtualBoxsoctalk-demo-0.2.1.vhdx.xz, Microsoft Hyper-V (Generation 1)soctalk-demo-0.2.1.vhd.xz, Microsoft Azure (fixed-size, 1 MiB aligned)soctalk-demo-0.2.1.raw.xz, generic cloud import (GCP, OpenStack),ddto physical diskSHA256SUMS.txt, checksums for all of the above
All disk images are compressed with xz -9. Decompress with xz -d <file>.xz.
The same release also carries a one-step VMware OVA and the native OS packages:
soctalk-demo-0.2.1.ova, VMware ESXi / vSphere one-step "Deploy OVF Template". See Run on VMware ESXisoctalk-0.2.1-1.x86_64.rpm·soctalk_0.2.1_amd64.deb, thesoctalkCLI + installer for dnf/apt hosts. See Install from an OS package
Format chooser
| Your platform | Download |
|---|---|
| Proxmox VE | qcow2, qm disk import. See Run on Proxmox |
| libvirt / virt-manager / QEMU CLI | qcow2 |
| KVM (RHEL/CentOS/Alma) | qcow2 |
| VMware ESXi | vmdk, see Run on VMware ESXi |
| VMware Workstation / Fusion | vmdk |
| VirtualBox | vmdk, convert to VDI, then attach. See Run on VirtualBox |
| Microsoft Hyper-V | vhdx, Generation 1 VM (validated; the image boots via BIOS firmware, Gen 2 / UEFI is untested) |
| Microsoft Azure | vhd, direct upload to a Managed Disk → Image → VM. See Run on Azure |
| Google Cloud | raw, tar czf disk.tar.gz disk.raw && gcloud compute images create |
| OpenStack | raw, openstack image create --disk-format raw |
| AWS | vmdk, import as an AMI with VM Import, or build a native AMI with Packer. See Run on AWS |
| Bare metal | raw, dd if=disk.raw of=/dev/sdX bs=4M |
Verify the download
curl -L -O https://github.com/soctalk/soctalk/releases/download/v0.2.1/SHA256SUMS.txt
curl -L -O https://github.com/soctalk/soctalk/releases/download/v0.2.1/soctalk-demo-0.2.1.qcow2.xz
sha256sum -c SHA256SUMS.txt --ignore-missingDisk size
All images have:
- Compressed size: ~600 MB to 1.5 GB depending on format
- Decompressed apparent size: 60 GB
- Actual on-disk size after first boot: ~3 GB (qcow2/vhdx are sparse; raw and vhd are fixed-size but most of the 60 GB is zeros and compresses to nothing)
Build it yourself
The Packer harness lives in infra/packer/. With KVM and Packer 1.11+ on a Linux host:
git clone https://github.com/soctalk/soctalk.git
cd soctalk/infra/packer
packer init .
packer build -only="soctalk-demo.qemu.soctalk_demo" .Builds in ~1 minute on a modern host with KVM acceleration. Outputs land in build/dist/: one file per format. See the Packer README for cloud-init seed configuration and the AWS AMI source.
CI workflow
build-packer-images.yml is workflow_dispatch-only; Packer builds aren't fired on every push because they're slow and consume runner minutes. Fire intentionally for a new release:
gh workflow run build-packer-images.yml -f version=0.2.1The workflow runs:
- build-wizard: Go build of the setup wizard binary.
- build-qcow2: Packer build producing all five formats; xz-compresses; uploads each as a separate workflow artifact; attaches to the GitHub Release on
v*tags. - boot-test: Boots the qcow2 in a fresh KVM VM, waits for
soctalk-firstbootto complete, runs a Playwright validator. Fails the workflow if the image is broken so a bad release never gets attached to a tag.
