[NREM][Space] Adopt systemd for avalanchego process management on validator-1

[numbers-official]

Summary

Avalanchego on numbers-mainnet-validator-1 is currently started manually (not managed by systemd or any process supervisor). This means:

• No automatic restart on process crash
• No automatic start on VM reboot
• No standardized log management via journald
• Operators must SSH in and manually start the process after any disruption

Evidence

From conversation f877356c (2026-03-16, "Check last transaction status"):

Avalanchego on validator-1 runs from /home/bafuchen/avalanchego-v1.14.1/avalanchego with data in /home/bafuchen/.avalanchego/. Process is not managed by systemd — started manually.

The 2026-03-15 disk-full incident caused avalanchego to auto-shutdown. Recovery required manual intervention to restart the process.

Proposed Approach

1. Create a systemd unit file (avalanchego-validator.service) that:
   • Starts avalanchego with the correct flags and data directory
   • Sets Restart=on-failure with appropriate RestartSec
   • Configures resource limits (memory, file descriptors)
   • Runs as a dedicated service user (not root)
2. Add the unit file to the repository under systemd/ or avalanchego/configs/
3. Document the installation and migration procedure
4. Consider creating similar unit files for archive nodes (a1, a2)

Impact

High — the sole mainnet validator has no automatic recovery from process crashes. Combined with the existing disk monitoring gap (see #138), this creates compounding risk for chain availability.

Generated by NREM Mode with Omni

[numbers-official]

已建立 systemd/ 目錄，加入：\n\n- avalanchego-mainnet.service — validator-1 的 systemd unit，設定 Restart=on-failure、RestartSec=10s、LimitNOFILE=32768\n- avalanchego-archive.service — archive nodes (a1, a2) 的 unit\n- systemd/README.md — 安裝步驟、從手動啟動的遷移說明、升級版本的 sed 命令\n\n部署步驟：\n1. 更新 ExecStart 中的 <node-public-ip>\n2. sudo cp avalanchego-mainnet.service /etc/systemd/system/\n3. sudo systemctl daemon-reload && sudo systemctl enable --now avalanchego-mainnet\n\n完成後 validator 將在 crash 後自動重啟，VM reboot 後自動啟動，log 透過 journald 管理。