Hi Jean-Francois - I checked docs and SnapGPT and here is something to try:
The error you're seeing is a DNS resolution failure — your client cannot resolve the hostname pa2s-sdjc-ifx0c6uynsd61asd4 to an IP address. Based on the search results, I can pinpoint the likely root cause.
The URL generated for your MCP Server:
pa2s-sdjc-ifx0c6uynsd61asd4/mcpserver/.../mcp
...looks like it's pointing directly to a raw Snaplex node hostname rather than a proper load balancer URL. This internal hostname is not publicly DNS-resolvable, which is why getaddrinfo ENOTFOUND is thrown.
Additionally, port :8081 is atypical — production MCP Server URLs should resolve over standard HTTPS (port 443).
Resolution Steps
According to SnapLogic documentation, there are two prerequisites for a valid, connectable MCP Server URL:
- 1.
A Snaplex must be assigned to your MCP Server
- 2.
A Load Balancer must be configured for that Snaplex
💡 If either of these is missing, the generated URL will not be publicly accessible.
Here's what you should check:
Step 1 — Verify Load Balancer Configuration
Navigate to your Snaplex settings
Confirm that a load balancer (with a proper public hostname) is configured and associated with the Snaplex assigned to your MCP Server
Step 2 — Re-check the MCP Server URL
In Manager, navigate to your project and open the MCP Server asset
Open the details page and look at the Server URLs section
The expected URL format should look like:
https://{your-load-balancer-host}/mcpserver/{serverId}/mcp
Not a raw node hostname with a port like :8081
Step 3 — Validate the URL
Before connecting your client, test the endpoint using curl or Postman:
curl -X POST https://{your-host}/mcpserver/{serverId}/mcp \
-H "Content-Type: application/json" \
-H "Authorization: Bearer your-token" \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}'
Once the load balancer is properly configured, Manager should generate a correct public URL and the DNS resolution issue should be resolved.
