Troubleshooting
This guide addresses high-frequency issues encountered during everyday platform operations, along with standard diagnostic procedures.
1. Scanning Probe (Client) Issues
1. Probe process running, but Web UI shows "Offline"
- Step 1: Check Client Secret
- The
-secretcommand-line flag passed totestnet-clientmust match the server'sTESTNET_CLIENT_SECRETenvironment variable character-for-character. Mismatched secrets cause the server to reject registration attempts.
- The
- Step 2: Test Server Reachability
- Run
curl -k -s -o /dev/null -w "%{http_code}\n" https://<server-ip>:3100on the probe node. A200response confirms network reachability and that the port is not blocked by firewall or security group rules (the default deployment publishes port 3100 only).
- Run
- Step 3: NTP Clock Synchronization
- Align probe and server hosts to a common time source (e.g.,
ntpdate pool.ntp.org) so log timelines can be cross-referenced and TLS certificate validation stays stable.
- Align probe and server hosts to a common time source (e.g.,
2. Probe is online, but workflow tasks remain "Queued"
- Verify Node Tool Whitelist:
- If an online node's tool whitelist does not include the tool used by the task, the dispatcher will not assign the task to it. Open "Scanning Nodes" → "Configure" to review the whitelist (empty means unrestricted).
- Concurrency Saturation:
- If active jobs consume all configured execution slots on a node, new jobs wait in queue until slots free up.
3. Docker Error: permission denied while trying to connect to the Docker daemon socket
- Cause: The OS user executing
testnet-clientlacks permissions to access/var/run/docker.sock. - Resolution:bash
sudo usermod -aG docker $USER newgrp docker # Or grant read/write access directly sudo chmod 666 /var/run/docker.sock
2. Workflows & Scan Task Issues
1. Workflow stalls at a specific node
- Check Upstream Output Volume:
- If an upstream stage (e.g., Subdomain Enumeration) yielded zero discovered targets, downstream stages (e.g., Web Probing) gracefully exit without error.
- Inspect Execution Logs:
- Open "Run Center", find the task, and click "View Logs" to inspect whether tools failed due to target DNS resolution failure or Out-Of-Memory (OOM) crashes.
2. Execution Error: Command not found / Tool not installed
- Native Shell Mode: Ensure binaries (e.g.,
subfinder,httpx) are installed and present in the system$PATH. - Docker Mode Recommendation: Prefer containerized tool definitions (e.g.,
registry.cn-hangzhou.aliyuncs.com/testnet-tools/subfinder:latest). Probes automatically pull images on first execution.
3. Asset Inventory Issues
1. Assets imported, but not appearing in the table
- Check Top Project Selector:
- TestNet enforces strict project isolation. Assets created or imported are pinned to the currently active project. Confirm that your active project matches the target workspace.
- Inspect Quick Filters:
- Check whether quick filters or advanced query conditions are active. Clear the active filters to expose all records.
2. Why wasn't a duplicate host created?
- Automated Deduplication: TestNet enforces strict deduplication rules:
- Root domains deduplicated by normalized lowercase FQDN;
- IP addresses deduplicated by IPv4/IPv6 canonical notation;
- Web applications deduplicated by standardized URI (scheme + host + port).
- Re-scanning an existing target refreshes the Last Seen timestamp and updates signatures without polluting data tables.
4. AI & MCP Integration Issues
1. Claude Desktop / Cursor Error: SSE connection failed (401 Unauthorized)
- Validate API Key:
- Verify that the
Authorization: Bearer <API_KEY>or query parameter matches the key shown in "AI" → "MCP Access".
- Verify that the
- Disable Proxy Buffering:
- When proxying SSE connections through Nginx, disable proxy buffering:nginx
proxy_set_header Connection ''; proxy_http_version 1.1; chunked_transfer_encoding off; proxy_buffering off; proxy_cache off; proxy_read_timeout 86400s;
- When proxying SSE connections through Nginx, disable proxy buffering:
5. Licensing Issues
1. "License Expired or Invalid Machine ID"
- Cause: Reinstalling the OS or altering network cards modifies the hardware fingerprint.
- Resolution:
- Run
./testnet-license show-machine-idon the server host to obtain the current fingerprint. - Request an updated
.licfile from your license issuer. - Upload the new file under "System" → "License Management"; the system validates and activates it immediately.
- Run
Related Documentation
- FAQ — quick answers to common questions
- Setup & Activation — installation and environment variables
- Node Security — probe sandbox and hardening