I am implementing a Facebook Registration Plugin. While in development locally and on one server, 开发者_Python百科it works perfectly. Once in production (6 webservers, behind a load balancer), it fails. It looks from the errors that we are getting (from our app) that the request the page is looking for is empty.
Our current theory is that the request is getting lost in the load balancer.
Does anyone have experience with this setup and can confirm or deny that this would be an issue?
It is a coyote point load balancer in front of 6 Apache boxes.
The reason it fails is due to the sessions saving. If you can keep session persistence on your load balancers, it should work as expected. Otherwise, you may need to build a work around, as I did at some point using cookies and various validation methods to keep the user's session persistent.
Found the solution - in this case, it was the load balancer, and was solved by making the redirect-uri value be set to the current server that the request was coming from. For instance, if the user was trying to register on server001, make the request-uri set to server001. This allowed the external request to make it to the right server, and the session took over the rest.
精彩评论