From 304198ebbdf2705e9b0383574a678b7797c711a8 Mon Sep 17 00:00:00 2001
From: alvis <allogn@gmail.com>
Date: Sun, 8 Mar 2026 06:51:40 +0000
Subject: [PATCH] Add GPU monitoring docs, Agap Installation page

---
 Agap-Installation.md | 85 ++++++++++++++++++++++++++++++++++++++++++++
 Home.md              |  1 +
 Zabbix.md            | 25 +++++++++++++
 3 files changed, 111 insertions(+)
 create mode 100644 Agap-Installation.md

diff --git a/Agap-Installation.md b/Agap-Installation.md
new file mode 100644
index 0000000..4bbebd5
--- /dev/null
+++ b/Agap-Installation.md
@@ -0,0 +1,85 @@
+# Agap Installation
+
+Steps to set up a fresh Agap server from scratch.
+
+## 1. GPU & Docker
+
+```bash
+sudo ./nvidia-docker-install.sh   # Docker + NVIDIA Container Toolkit
+./install-cuda.sh                  # CUDA toolkit (no driver)
+```
+
+## 2. Zabbix Agent (host)
+
+Install agent and plugins:
+
+```bash
+# Add Zabbix repo
+wget https://repo.zabbix.com/zabbix/7.4/release/ubuntu/pool/main/z/zabbix-release/zabbix-release_latest_7.4+ubuntu24.04_all.deb
+sudo dpkg -i zabbix-release_latest_7.4+ubuntu24.04_all.deb
+sudo apt update
+
+# Install agent and GPU plugin
+sudo apt install zabbix-agent2 zabbix-agent2-plugin-nvidia-gpu
+```
+
+Configure `/etc/zabbix/zabbix_agent2.conf`:
+
+```ini
+Server=127.0.0.1
+ServerActive=127.0.0.1:10051
+Hostname=AgapHost
+PluginSocket=/run/zabbix/agent.plugin.sock
+ControlSocket=/run/zabbix/agent.sock
+Include=/etc/zabbix/zabbix_agent2.d/plugins.d/*.conf
+Include=/etc/zabbix/zabbix_agent2.d/*.conf
+```
+
+```bash
+sudo systemctl enable --now zabbix-agent2
+```
+
+In Zabbix UI, link these templates to the `AgapHost` host:
+- **Linux by Zabbix agent active**
+- **Nvidia by Zabbix agent 2 active**
+
+## 3. Custom Zabbix UserParameters
+
+Add backup monitoring to `/etc/zabbix/zabbix_agent2.d/gitea_backup.conf`:
+
+```ini
+# Gitea backup
+UserParameter=gitea.backup.status,grep -c "Finish dumping" /mnt/backups/gitea/backup.log 2>/dev/null | grep -qx 1 && echo 1 || echo 0
+UserParameter=gitea.backup.age,f=$(ls -t /mnt/backups/gitea/gitea-dump-*.zip 2>/dev/null | head -1); [ -n "$f" ] && echo $(( $(date +%s) - $(stat -c %Y "$f") )) || echo -1
+
+# DBS backup
+UserParameter=dbs.backup.age,f=/mnt/backups/dbs/.last_sync; [ -f "$f" ] && echo $(( $(date +%s) - $(stat -c %Y "$f") )) || echo -1
+
+# Immich backup
+UserParameter=immich.backup.age,f=/mnt/backups/media/.last_sync; [ -f "$f" ] && echo $(( $(date +%s) - $(stat -c %Y "$f") )) || echo -1
+```
+
+## 4. Root Cron Jobs
+
+```bash
+sudo crontab -e
+```
+
+Add:
+```
+0 3 * * * /home/alvis/agap_git/gitea/backup.sh >> /mnt/backups/gitea/cron.log 2>&1
+30 2 * * * /home/alvis/agap_git/immich-app/backup.sh >> /mnt/backups/media/cron.log 2>&1
+30 3 * * * rsync -a --delete /mnt/ssd/dbs/ /mnt/backups/dbs/ >> /mnt/backups/dbs/cron.log 2>&1 && touch /mnt/backups/dbs/.last_sync
+```
+
+## 5. Services
+
+Start all Docker services:
+
+```bash
+cd ~/agap_git
+docker compose up -d                        # Immich
+cd gitea && docker compose up -d            # Gitea
+cd ../openai && docker compose up -d        # Open WebUI + Ollama
+cd ../zabbix && docker compose up -d        # Zabbix
+```
diff --git a/Home.md b/Home.md
index 78e010f..2bba45f 100644
--- a/Home.md
+++ b/Home.md
@@ -2,6 +2,7 @@
 
 ## Infrastructure
 
+- [[Agap-Installation]] — Fresh install guide
 - [[Network]] — Netplan, Caddy, port forwarding
 - [[Storage]] — LVM setup
 - [[Backups]] — Gitea and database backups
diff --git a/Zabbix.md b/Zabbix.md
index b9a283d..8e53daf 100644
--- a/Zabbix.md
+++ b/Zabbix.md
@@ -64,6 +64,31 @@ Host "HA Agap" receives alerts from Home Assistant via `history.push` API.
 
 To add a new HA alert: create a trapper item + trigger on "HA Agap", add `rest_command` in HA `configuration.yaml`, create HA automation to call it.
 
+## GPU Monitoring (AgapHost)
+
+The host agent monitors the GTX 1070 GPU via the `zabbix-agent2-plugin-nvidia-gpu` package.
+
+**Installed packages:**
+```bash
+apt install zabbix-agent2-plugin-nvidia-gpu
+```
+
+Plugin binary: `/usr/libexec/zabbix/zabbix-agent2-plugin-nvidia-gpu`
+Plugin config: `/etc/zabbix/zabbix_agent2.d/plugins.d/nvidia.conf`
+
+Template linked to AgapHost: **Nvidia by Zabbix agent 2 active** — uses `nvml.*` keys.
+
+Key metrics reported:
+| Item | Key |
+|------|-----|
+| GPU memory free | `nvml.device.memory.fb.free` |
+| GPU memory used | `nvml.device.memory.fb.used` |
+| GPU utilization | `nvml.device.utilization.gpu` |
+| Temperature | `nvml.device.temperature` |
+| Power usage | `nvml.device.power.usage` |
+
+> ECC memory error items show expected errors — GTX 1070 does not support ECC.
+
 ## Notes
 
 - Zabbix server port 10051 is exposed on the host for the host agent