Azure Machine Learning HTTP Error 500: Internal Server Error In Private Endpoints

Question

Azure Machine Learning HTTP Error 500: Internal Server Error . Is this feature no longer available? Is this a backend or platform error?

Answer 1

Hi ZTS

Below is step-by-step approach used in debugging workspace connectivity issue

Step 0: Check for ongoing outages or Service health issue in your subscription

Step 1: Network trace

Take a network trace while accessing workspace or workspace resources to see blocked or failing URLS. (403, 404 or 500's)

and follow step 2 to see whether respective URL resource is properly secured against private endpoint or not.

Step 2: DNS resolution check

Please go through all workspace resource to see if all are properly secured by creating private endpoint against virtual network as mentioned in this document.

Check DNS resolution of each workspace resource whether they are resolving to private IPs as mentioned in in DNS configuration or not.

nslookup <fqdnofresourceasmentioned in DNS record>

Please make sure you have disable private endpoint policy while creating the private endpoint

If the DNS are not resolving

in Azure DNS scenario, you can flush old DNS record and create new ones as per DNS configuration or re-create the private endpoints after deleting the old ones

in Custom DNS scenario, please make sure you have added the A-records in host server as per DNS configuration.

Step 3: Whitelisting in Org firewall or proxy or security software side

If DNS is resolving, then please check your org firewall/proxy.

Step 4: Outbound traffic in Network security group (NSG)

You have allowed necessary service tags in NSG for outbound communication

Reference on modifying outbound traffic in NSG

Step 4: Adjust TTL setting in private endpoint DNS to attain higher DNS queries

Reference on Workspace connectivity Troubleshooting

Reference on Securing Workspace Resources

Hope it helps.

Thank you