2 Content-Type values for the ajax call

[Chan Chea Hang]

Hey guys,

I've been having some trouble with OpenLiteSpeed 1.4.23 Ajax calls returning 2 Content-Type values which causes my WHMCS client area to go hay-wire. I've also contacted WHMCS support in regards to this issue and they have recommended me to contact Litespeed in regards to the case as Apache isn't being affected this issue.

The expected Content-Type value in the ajax call response headers are:

Content-Type:application/json

However your server is returning:

content-type:text/html; charset=utf-8
content-type:application/json

Another user reported the same problem:
https://www.litespeedtech.com/support/forum/threads/litespeed-issue-with-json.6217/

Is there a way to stop Litespeed from serving the ajax content with the text/html content-type?

Thanks so much for any help and responses.

 

[mistwang]

We need to find out how content-type: text/html get there.
The ajax request is served by PHP?

 

[mistwang]

If you build OLS from source, you can try the following patch

diff --git src/http/httpcgitool.cpp src/http/httpcgitool.cpp
index b7b22e6..cf82585 100644
--- src/http/httpcgitool.cpp
+++ src/http/httpcgitool.cpp
@@ -137,7 +137,7 @@ int HttpCgiTool::processHeaderLine(HttpExtConnector *pExtConn,
             str.append(pLineBegin, pLineEnd - pLineBegin);
             str.append(pCharset->c_str(), pCharset->len());
             str.append("\r\n", 2);
-            buf.parseAdd(str.c_str(), str.len(), LSI_HEADEROP_ADD);
+            buf.parseAdd(str.c_str(), str.len());
         }
         return 0;
     case HttpRespHeaders::H_CONTENT_ENCODING:

 

[Chan Chea Hang]

We need to find out how content-type: text/html get there.
The ajax request is served by PHP?

Highly likely. The request is sent to eNom's API (https://www.enom.com/api/) and received on WHMCS client area (mainly affecting my domain whois lookup and is developed based on PHP)

 

[Chan Chea Hang]

Update: I've installed OLP from source with the above-mentioned patch. It's still serving the headers with 2 content-type.

Here's the headers as captured by Fiddler:

Damn... Any other possible solutions?

 

[mistwang]

Do you have a test script that we can reproduce this problem. if we can reproduce it, we will fix it easily.