Wednesday, April 15, 2026

New Computer

______________________________________

CERBERUS REBUILD CONTEXT (PASTE THIS)

I have reinstalled CERBERUS (Celeron NUC) and need to reconnect it to my existing system.
System layout

    ANNIE (192.168.50.172) = main machine (client)
    CERBERUS (192.168.50.10) = storage server (NFS)
    Two drives:
        STORAGE (primary)
        ARCHIVE (secondary)

Goal

    Mount both drives on CERBERUS at:
        /srv/STORAGE
        /srv/ARCHIVE
    Export them via NFS to ANNIE
    ANNIE mounts them at:
        /mnt/STORAGE
        /mnt/ARCHIVE

Known disk info (from previous system)

    STORAGE UUID: 9f9c11ef-318b-4d39-b544-8fef87049095
    ARCHIVE UUID: 76ad1e62-25d3-4bf7-840d-aa6c09658e7a

Expected CERBERUS config

/etc/fstab

UUID=9f9c11ef-318b-4d39-b544-8fef87049095 /srv/STORAGE ext4 defaults 0 2
UUID=76ad1e62-25d3-4bf7-840d-aa6c09658e7a /srv/ARCHIVE ext4 defaults 0 2

/etc/exports

/srv/STORAGE 192.168.50.172(rw,sync,no_subtree_check)
/srv/ARCHIVE 192.168.50.172(rw,sync,no_subtree_check)

Requirements

    Use stable mount points (/srv/...)
    Use labels or UUIDs (no /media/...)
    Keep it minimal, no desktop auto-mounting

What I need

Walk me step-by-step to:

    Mount drives correctly
    Set up fstab
    Install + configure NFS
    Verify from ANNIE

What this gives you

When you paste that:

    I instantly know your topology ✅
    I know your disk IDs ✅
    I know your mount philosophy ✅
    We skip all exploration ✅

Optional (keep with your backup)

You already saved:

/mnt/STORAGE/system-backup/cerberus-rebuild.txt

That file = ground truth

If you want next:

    I can give you a 10-minute rebuild checklist
    or move into rsync cleanup plan

Credits Used: 15.8

 .......................................................................................

since we have made those good nano changes on cerberous ... and will be doing a clean install later it seems a shame to have to do that again .. should we save those bits of the setup .. ?
RouteLLM
Routing to GPT-5.3 Instant

Yes — and this is exactly the right move.

You don’t want to remember this — you want to replay it.
✅ What to save (minimal but complete)

You only need to preserve 3 things from CERBERUS:
1. Mount logic (fstab)

bash
Copy
cat /etc/fstab

2. NFS export config

bash
Copy
cat /etc/exports

3. Drive identity (for sanity later)

bash
Copy
lsblk -f

✅ Save them to STORAGE (so they survive reinstall)

On CERBERUS:

bash
Copy
mkdir -p /srv/STORAGE/system-backup

Then:

bash
Copy
cat /etc/fstab > /srv/STORAGE/system-backup/cerberus-fstab.txt
cat /etc/exports > /srv/STORAGE/system-backup/cerberus-exports.txt
lsblk -f > /srv/STORAGE/system-backup/cerberus-disks.txt

✅ Optional (very useful)

Also save:

bash
Copy
hostnamectl > /srv/STORAGE/system-backup/cerberus-system.txt
ip a > /srv/STORAGE/system-backup/cerberus-network.txt

🧠 Why this works

After reinstall you won’t:

    guess UUIDs ❌
    guess mount points ❌
    guess exports ❌

You just:

    reinstall OS
    open these files
    copy back into place

Done in minutes.
✅ Even better (one-file version)

If you want a single reference:

bash
Copy
(
echo "=== FSTAB ==="
cat /etc/fstab
echo
echo "=== EXPORTS ==="
cat /etc/exports
echo
echo "=== DISKS ==="
lsblk -f
) > /srv/STORAGE/system-backup/cerberus-rebuild.txt

✅ Result

You now have a rebuild blueprint.

👉 Do you want next:

    a post-reinstall checklist (5 commands to restore everything)
    or move into rsync cleanup plan

Credits Used: 13.49 

...............................................

 

📔 Master Field Notes: The NUC Mesh System

Version: 1.0
Machines: ANNIE (i7 NUC9), CERBERUS (Celeron NUC), PI-GATEWAY (Pi 5)
1. System Architecture: The "Distributed Cluster"

Instead of thinking of these as three separate PCs, they function as a single unit:

    ANNIE (The Cockpit): Intel NUC9i7QNX. The "Power Engine." Handles all visuals and the primary user interface.
    CERBERUS (The Storage): Celeron NUC. The "Limb." Tucked away, dedicated to the 2 x HDD movie/file archive.
    PI-GATEWAY (The Nerve Center): Raspberry Pi 5. Handles DHCP and internet gateway duties.

2. Graphics Stability (Primary Fix)

ANNIE initially suffered from "Jumping Windows" and transparency glitches due to a conflict between the NVIDIA driver and the XFCE compositor.

    The Philosophy: "Horses that eat must pull." No-frills reliability over fancy effects.
    The Fix: Disable the Compositor permanently.
        Manual recovery: xfwm4 --replace --compositor=off &
        Permanent Fix: Settings -> Window Manager Tweaks -> Compositor -> Untick "Enable display compositing".
    Hardware Profile:
        GPU: NVIDIA GTX 1650 (ASUSTeK).
        Driver: Proprietary NVIDIA (Loaded via ubuntu-drivers autoinstall).
        Monitor Logic: Dedicated GPU ports only. Use a $10 "DisplayPort to HDMI" adapter for secondary screens.

3. Remote Management (The "Mirror")

We use a Cockpit & Broadcaster model to eliminate the need for three keyboards/mice.

    Protocol: VNC (Port 5900).
    Setup on Broadcaster (CERBERUS):
        Tool: x11vnc (shares the actual physical screen :0).
        Set Password: x11vnc -storepasswd
        Broadcast Command: x11vnc -auth guess -forever -loop -noxdamage -rfbauth ~/.vnc/passwd -shared &
    Setup on Cockpit (ANNIE):
        Tool: Remmina with VNC plugin.
        Usage: Open Remmina, select VNC, enter IP (192.168.50.10), and save credentials.

4. The SSH Mesh (The "Nervous System")

All machines are linked via SSH Keys for password-less, encrypted communication.

    ANNIE IP: 192.168.50.172
    CERBERUS IP: 192.168.50.10
    Key Config: Created ~/.ssh/config on all machines to use aliases (e.g., typing ssh cerberus instead of the full IP).

5. Hardware Manifest (ANNIE)

Keep this for driver/kernel troubleshooting:

    Kernel: 6.17.0-20-generic (Linux Mint 22.3 Zena / Ubuntu 24.04 base).
    CPU: Intel Core i7-9750H (6-core/12-thread).
    Memory: 16GB.
    OS Drive: 1TB Samsung NVMe SSD 970 EVO.
    Network: Intel Wi-Fi 6 AX200 & Dual Intel Gigabit Ethernet.

6. Maintenance Rule #1

Timeshift is the Insurance Policy.
Before any major "maintenance" or "fiddling," run a Timeshift snapshot. If the windows start jumping again or the session breaks, roll back to the "Post-Driver/No-Compositor" snapshot.