Commit Graph
2 Commits
Author SHA1 Message Date
beatzaplentyandClaude Sonnet 4.6 23634134f0 fix(ha-test): remove Debian LVM handlers and fix corosync authkey length
- Remove before/after-resync-target LVM handlers (snapshot-resync-target-lvm.sh
  doesn't exist in NixOS; exit code 127 caused DRBD to drop connections on sync)
- Remove split-brain handler pointing to /usr/lib/drbd/ (Debian path)
- Fix testAuthKey from 126 to 128 bytes (corosync minimum is 1024 bits)
- Fix cluster-init.sh quorum check: corosync-quorumtool has no -q flag;
  use -s | grep 'Quorate: Yes' instead

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HaH1cSGvhogRP5ExoF6nD8
2026-07-27 06:40:57 +10:00
beatzaplentyandClaude Sonnet 4.6 750121e9dd test-lab: add two-node HA file-server test cluster config
Disposable test VMs (ha-test-node1 / ha-test-node2, VMIDs 200/201 on pve1)
to evaluate whether the DRBD + XFS + LIO + Corosync + Pacemaker stack
runs correctly on NixOS before deciding NixOS vs Debian for production.

Includes:
- test-lab/ha/disko.nix: 20G boot disk layout (smaller than production)
- test-lab/ha/common.nix: shared HA stack (drbd, corosync, pacemaker,
  targetcli-fb, xfsprogs), OCF PATH workaround for nixpkgs#207891
- test-lab/ha/node1.nix / node2.nix: per-node hostname + static IP
- test-lab/ha/fence-pve-ssh.py: Proxmox SSH fence agent for STONITH
- test-lab/ha/cluster-init.sh: one-shot cluster bootstrap script
- test-lab/ha/cluster-enable-stonith.sh: enables STONITH post-key-deploy
- flake.nix: adds ha-test-node1 / ha-test-node2 nixosConfigurations
  (bypasses mkTarget / clan-core / sops-nix — test-only)

These VMs must be destroyed once acceptance testing is complete.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-07-27 05:46:04 +10:00