babf062d6a
Original planning folder (no git repo) for the server diagnostics system that runs on CT 300. Live deployment is on claude-runner; this preserves the Agent SDK reference, PRD with Phase 2/3 roadmap, and N8N workflow designs. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
140 lines
3.9 KiB
Markdown
140 lines
3.9 KiB
Markdown
# N8N Setup Instructions
|
||
|
||
This guide walks through configuring the N8N health check workflow.
|
||
|
||
## Prerequisites
|
||
|
||
- N8N running at http://10.10.0.210:5678
|
||
- SSH access configured between N8N LXC and Claude Code LXC (already done)
|
||
- Discord webhook configured (already done)
|
||
|
||
## Step 1: Create SSH Credential in N8N
|
||
|
||
1. Open N8N at http://10.10.0.210:5678
|
||
2. Go to **Settings** (gear icon) → **Credentials**
|
||
3. Click **Add Credential**
|
||
4. Search for and select **SSH**
|
||
5. Configure:
|
||
- **Credential Name:** `Claude Code LXC`
|
||
- **Host:** `10.10.0.148`
|
||
- **Port:** `22`
|
||
- **Username:** `root`
|
||
- **Authentication:** `Private Key`
|
||
- **Private Key:** Paste the following:
|
||
|
||
```
|
||
-----BEGIN OPENSSH PRIVATE KEY-----
|
||
b3BlbnNzaC1rZXktdjEAAAAABG5vbmUAAAAEbm9uZQAAAAAAAAABAAAAMwAAAAtzc2gtZW
|
||
QyNTUxOQAAACAtHWX+TOsNkiL7TP1niwpO5AJjBThghODpa1q93g7bagAAAJiq1/Cyqtfw
|
||
sgAAAAtzc2gtZWQyNTUxOQAAACAtHWX+TOsNkiL7TP1niwpO5AJjBThghODpa1q93g7bag
|
||
AAAEBVXct4LSCsRexl9JYZtrYx4YVoioYBdWHlPinW+PudBS0dZf5M6w2SIvtM/WeLCk7k
|
||
AmMFOGCE4OlrWr3eDttqAAAAEm44bi10by1jbGF1ZGUtY29kZQECAw==
|
||
-----END OPENSSH PRIVATE KEY-----
|
||
```
|
||
|
||
6. Click **Save**
|
||
|
||
## Step 2: Import the Workflow
|
||
|
||
1. Go to **Workflows** in N8N
|
||
2. Click **Add Workflow** → **Import from File**
|
||
3. Select: `/mnt/NV2/Development/headless-claude/n8n-workflow-import.json`
|
||
- Or copy the workflow from this repo
|
||
4. After import, open the workflow
|
||
|
||
## Step 3: Link the SSH Credential
|
||
|
||
1. Click on the **Run Claude Diagnostics** node
|
||
2. In the **Credential** dropdown, select `Claude Code LXC`
|
||
3. Click **Save**
|
||
|
||
## Step 4: Test the Workflow
|
||
|
||
1. Click **Execute Workflow** (play button)
|
||
2. Watch each node execute:
|
||
- **Every 5 Minutes** → triggers
|
||
- **Run Claude Diagnostics** → SSHs to Claude Code LXC
|
||
- **Parse Claude Response** → extracts result
|
||
- **Has Issues?** → routes based on health
|
||
- **Discord Alert** or **Discord OK** → sends notification
|
||
|
||
3. Check Discord for the notification
|
||
|
||
## Step 5: Activate the Workflow
|
||
|
||
1. Toggle the workflow to **Active** (top right)
|
||
2. The workflow will now run every 5 minutes automatically
|
||
|
||
## Workflow Behavior
|
||
|
||
### Schedule
|
||
- Runs every 5 minutes by default
|
||
- Adjust in "Every 5 Minutes" node settings
|
||
|
||
### Health Check
|
||
- Calls Claude Code headless mode
|
||
- Runs `python3 ~/.claude/skills/server-diagnostics/client.py health paper-dynasty`
|
||
- Claude analyzes results and summarizes
|
||
|
||
### Notifications
|
||
- **Issues Found**: Red Discord embed with alert details
|
||
- **All Healthy**: Green Discord embed (disabled by default - enable "Discord OK" node if wanted)
|
||
|
||
### Cost
|
||
- Each health check costs ~$0.08 (Claude API usage)
|
||
- At 5-minute intervals: ~$0.08 × 12 × 24 = ~$23/day
|
||
- Consider increasing interval to 15-30 minutes for cost savings
|
||
|
||
## Customization
|
||
|
||
### Change Check Interval
|
||
1. Open "Every 5 Minutes" node
|
||
2. Adjust `minutesInterval` value
|
||
|
||
### Add More Servers
|
||
1. Update config.yaml on Claude Code LXC
|
||
2. Modify the Claude prompt in "Run Claude Diagnostics" node
|
||
|
||
### Enable OK Notifications
|
||
1. Click "Discord OK (Optional)" node
|
||
2. Toggle off "Disabled" setting
|
||
|
||
## Troubleshooting
|
||
|
||
### SSH Connection Failed
|
||
```bash
|
||
# Test from N8N LXC
|
||
ssh -i ~/.ssh/n8n_to_claude root@10.10.0.148 "hostname"
|
||
```
|
||
|
||
### Claude Command Not Found
|
||
```bash
|
||
# Verify Claude path on Claude Code LXC
|
||
/root/.local/bin/claude --version
|
||
```
|
||
|
||
### Authentication Expired
|
||
If Claude returns auth errors, SSH to Claude Code LXC and re-authenticate:
|
||
```bash
|
||
ssh root@10.10.0.148
|
||
/root/.local/bin/claude # Follow device code flow
|
||
```
|
||
|
||
## Architecture
|
||
|
||
```
|
||
N8N (10.10.0.210)
|
||
│
|
||
├── Schedule Trigger (every 5 min)
|
||
│
|
||
└── SSH to Claude Code LXC (10.10.0.148)
|
||
│
|
||
└── Claude Code Headless Mode
|
||
│
|
||
└── server-diagnostics skill
|
||
│
|
||
└── SSH to Paper Dynasty (10.10.0.88)
|
||
│
|
||
└── Docker health checks
|
||
```
|