cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
618
Views
0
Helpful
2
Replies

http-bind POST becomes a GET?

I am troubleshooting an intermittent (but frequent) connection failure and recently saw Chrome dev tools network trace that showed the http-bind request to port 7443 was sent as a GET and not a POST.  The server then returned a 400 Bad Request, of course.

Is there any reason that the http-bind would be sent as a GET and not a POST? I checked in a known working agent and 100% of the http-bind requests were POSTs.

I know that this isn't the root cause of connection failures, but it feels like a promising thread to pull on. If the network were interrupted in some way just prior to the normal POST, would Finesse change to using GET? I cannot imagine any logic in the code that would alternate POST with GET if the expected result would be HTTP 400! Is it possible the browser misread the http method from its trace somehow? 

2 Replies 2

dekwan
Cisco Employee
Cisco Employee

Hi,

 

Are you seeing this issue with the Finesse out of the box desktop? Or is this a custom desktop?

 

I have not dug deep into network issues with the network trace so I don't know if the bind gets sent as a GET during a network interruption, but as you said, I can't image that logic being there.

 

Let's take a step back and let me ask what is the issue you are trying to debug when you discovered this?

 

Thanx,

Denise

I'm sorry I didn't follow up on this - I was moved off the issue team to work on new development. I researched and the likely cause of a POST turning into a GET is a redirect, but the cause (or source) of the redirect is not clear. 

What happens is that Finesse (OOB) loses its BOSH connection in several different ways (it's not always the POST->GET scenario). I believe it's a networking or environmental issue. You know the old rule: if the same software both works and fails, the failure's trigger is not inside the software. It's reliably broken for some users, and just reliable for others.

Anyway, I'm no longer up on the current state of the issue, I just know it's still happening and we're moving to 12.6 to mitigate.