Proxy Server Tool (proxy.jar)

The proxy server acts as an intermediate server that relays requests between a client and a server. The proxy server keeps track of all the client-server interactions and outputs a log of the entire TCP communication. This allows you to monitor exactly what is going on, without having to access the main server.

You can find the proxy server in the appropriate installation folder:

• <cq_install_path>/opt/helpers/proxy.jar
• <crx_install_path>/opt/helpers/proxy.jar

You can use the proxy server to monitor all client-server interaction, regardless of the underlying communication protocol. For example, you can monitor the following protocols:

• HTTP for Web pages
• HTTPS for secure Web pages
• SMTP for email messages
• LDAP for user management

For example, you can position the proxy server between any two applications that communicate via a TCP/IP network; e.g. a web browser and AEM. This allows you to monitor exactly what happens when you request a AEM page.

The tool can be found in the /opt/helpers folder of your AEM installation. To start it type:

• q (quiet Mode) Does not write the requests to the console window. Use this if you do not want to slow down the connection, or if you log the output to a file (see -logfile option).
• b (binary Mode) If you are looking for specific byte combinations in the traffic, enable binary mode. The output will then contain the hexadecimal and character output.
• t (time stamp log entries) Adds a time stamp to each log output. The time stamp is in seconds, so it may not be suitable for checking single requests. Use it to locate events that occurred at a specific time if you use the proxy server over a longer time period.
• logfile <filename> (write to log file) Writes the client-server conversation to a log file. This parameter works also in quiet mode.
• i <numIndentions> (add indention) Each active connection is indented for better readability. Default is 16 levels. (New in proxy.jar version 1.16).

The following scenarios illustrate a few of the purposes for which the Proxy Server Tool can be used:

Check for Cookies and their Values

The following log entry example shows all cookies and their values sent by the client on the 6th connection opened since proxy start:

Checking for Headers and their Values The following log entry example shows that the server is able to make a keep-alive connection and the content length header was properly set:

Checking if Keep-Alive works

Keep-Alive means that a client re-uses the connection to the server to transports multiple files (the page code, pictures, style sheets and so on). Without keep-alive, the client has to establish a new connection for each request.

To check if keep-alive works:

1. Start the proxy server.
2. Request a page.

• If keep-alive is working, the connection counter should never go above 5 to 10 connections.
• If keep-alive is not working, the connection counter increases rapidly.

Finding Lost Requests

If you lose requests in a complex server setting, for example with a firewall and a dispatcher, you can use the proxy server to find out where the request was lost. In case of a firewall:

1. Start a proxy before a firewall
2. Start another proxy after a firewall
3. Use these to see how far the requests are getting.

Requests Hanging

If you experience hanging requests from time to time:

1. Start a proxy.jar.
2. Wait or write the access log into a file - with each entry having a timestamp.
3. When the request start hanging you can see how many connections were open and which request is causing trouble.

The log entries produced by proxy.jar all have the following format:

For example, a request for a Web page may look as follows:

• C signifies that this entry comes from the client (it is a request for a Web page)
• 0 is the connection number (the connection counter starts at 0)
• #00000 the offset in the byte stream. This is the first entry, so the offset is 0.
• [GET <?>] is the content of the request, in the example one of the HTTP headers (url).

When a connection closes, the following information is logged:

This shows the number of bytes that passed between client and server on the 6th connection and at the average speed.

We shall review a simple template which produces the following code when requested:

If AEM is running on localhost:4303, start the proxy server as following:

You can access the server (localhost:4303) without the proxy server, but if you access it via localhost:4444, the proxy server will log the communication. Open a browser and access a page created with the above template. After that, look at the log file.

When starting, the following information is written to the log:

The following header fields are listed at the start of the first connection (0), which is requesting the main HTML page:

The Client requests a keep-alive connection, so the server can send multiple files over the same connection:

The proxy server is a good tool to verify whether cookies are properly set or not. Here, we see the:

• cq3session cookie generated by AEM
• the show mode switch cookie generated by the CFC
• a cookie named JSESSIONID; this is automatically created by JSP if not explicitly switched off using <%@ page session="false" %>:

The server will close connection 0 after the request. Keep-alive is not possible, because the request has a question mark. This means that the server cannot return a cached version, and therefore cannot determine the content length at this point, which is required for a keep-alive connection.

Here, the server starts sending the HTML code on connection 0:

Connection 0 closes immediately after the HTML file has been served:

Now, the output starts for connection 1, which downloads the image contained in the HTML code:

Again, the client requests a keep-alive connection:

For connection 1, the server can provide keep-alive, because the image is static and therefore the content length is known.

The server returns the content length of the image on connection 1:

Now that the content length is established, the server sends the image data on connection 1:

After the keep-alive timeout has been reached, connection 1 closes as well:

The above example is comparatively simple, because the two connections occur sequentially:

• first the server returns the HTML code
• then the browser requests the image and opens a new connection

In practice, a page may generate many parallel requests for images, style sheets, JavaScript files, etc. This means that the logs have overlapping entries of parallel open connections. In that case, we recommended to use option -i to improve readability.