Discourse load balancer encounters problems, hoping to help

After the discourse configuration is configured with a load balancing, there is a problem with the access. There is only one white board page. The load balancing rear server configuration can be normal. When the load balanor is configured with two back -end servers, it is strange to find a problem. Any help will be grateful

How did you install the individual sites? Did you get them to work without the list balancer?

Look at the javascript console for errors.

The two discourse sites are connected to the same postgreSQL and Redis database respectively, forwarded to port 80 through the nginx.sock file, and then the 80 port load of two machines is performed through the AWS AlB load balancer.

How did you launch them? Did this ever work?

Did you look in the javascript console for errors? I suspect it’ll show you that assets aren’t being loaded.

The current architecture is AWS EC2 responsible for the startup Discourse application. AWS PostgresQL and AWS Redis are responsible for data storage and application cache processing. The loads of two EC2 servers are performed through the AWS ALB load average, and ALB accelerates through AWS CloudFront; Well, ok, wipe
Try to bind the cloud layer to Alb Load Balanner through multiple attempts, and it is abnormal when accessing the page. How can I solve this situation? Very strange, you can access the AWS global accelerator normally, but when there is no doubt that when Cloudfront will be better than GA, I hope to get help, thank you

Please do not open multiple topics about the same issue. I have moved your post to the existing topic.


A single ec2 with cloudfront as cdn (Enable a CDN for your Discourse) is likely what I would recommend knowing what I do now (which might not be everything).

What level of traffic are you expecting?

1 Like

According to the expected traffic, it may reach 100000