Hello @Ashwani Tiwari , Thank you for reaching out.
I think this issue might be due to health probe load balancing setting or maybe due to session affinity. As documented here Front Door selects a backend to use within the backend pool. Backend selection is based on several factors, including:
- The health of each origin, which Front Door monitors by using health probes.
- The routing method for your origin group.
- Whether you have enabled session affinity
Front Door determines backend health as per the Load balancing settings such as Sample size
and Successful sample size
configured while adding the backend pool. This is the logic on how Front Door determines backend health. You can change these settings and then see if you observe the same issue
Another setting you can check is if session affinity is not enabled. This setting was mentioned in step 5 here from the tutorial.
You can also validate if caching is not enabled for any of the routing rules.
Hope this helps! Please let me know if the issue still persists. Thank you!