tcpserver | a very simple & convenient tcp server framework | TCP library

The error is telling us:

in the TCPServer Execute method there is no need to use a try - catch block. But what happens if in the TCPServer Execute method another kind of exception occurs?

ANY uncaught exception that is allowed to escape from the OnConnect or OnExecute event back into TIdTCPServer will cause the calling TIdContext thread to stop running. It will close its associated client Connection during its cleanup, firing the OnDisconnect event. And then the OnException event will be fired afterwards.

In my server application, I have put a TIdTCPServer component on the main form, if something bad happens inside its Execute method and I don't manage it, the server stops.

The server as a whole does not stop. Only the calling TIdContext thread is stopped.

Obviously I need the server running so I use a try - catch block and if there is an exception (any kind of exception), I restart the server.

It is not necessary to restart the whole server on any exception. Only on exceptions that invalidate/corrupt something that your app needs to function properly.

Sometimes, with the server application running in the IDE, I got the error

First, the default Windows Indy socket buffer is 32 KB big in size. So, when You would like to send all in one (what I do not prefered) make Your sending packets smaller. So, You can quicker check Your connection for Time-Out's or other disadvantages, too. BTW Question: what Version of Indy do You use - Version 9 or 10 ?

Second, check Your server/client Codes where You read/write the "request", and "responses" exists - A read function should end with \r\n when You in text mode.

Third, check, if You have non-blocking or blocking server Code. This means - You should install a TThread for each Connection to handle all Connection's properly (may be with other properties). So, the Connections do not overlap existing previous Connections.

Fourth (for advanced developers): You should be consider, to have a "second" Connection to each one - this can help to hold the line (hops...

Fiveth: (for advanced develppers): You should be in mind with security aspects. This means, You should using SSL certificates for Your Connections (Indy provides simple click, and edit components for this. You task is it, to create a "self signed certificate" (PuttyGen (on Windows) or on Linux - see google for details) - if You own Your public certificate, You don't need this step - have a look to "Lets Encrypt" (a public free SSL authority, but be warned: You have to renew develper SSL certificates each third month (a certbot script helps You to do all these things).

If You have question's, feel free to ask, I will try to help You.

Happy X-mas days

You have issues in both the client and server.

The client uses socket .recv() to read data but this will block forever at the end of the first command when the server output finishes. it will return b'' only at EOF, when the socket is closed by the server. Because of the server issue (below), it SEEMS to work.

Your server code only reads one line, spawns the command, sends the output and then returns out of handle() after which the connection is closed. Because of this close, the client actually works (it gets b'' from recv()).

To fix the server, do readline() in a loop until it returns b'' (socket closed by client) and preferably flush() the write stream after sending the command output.

Example server handler:

recv() stores only bytes that were actually received into the receiving buffer. In particular, it does not add a string terminator after the bytes it actually receives. If you want to interpret the data as a C string, then it is your responsibility to either

• add one, or
• ensure that the server sends one, and that the client doesn't consider the transmission complete until it receives that.

So,

is recvline[MAXLINE] combining the two together?

Effectively, yes. The characters 'y', 'e', 'l', 'l', 'o', 'w' are recorded in that array by the first recv() call. You are lucking out (or maybe unlucking out) that the next byte happens to be a null character, so that when you print the contents of recvline, you get exactly "yellow". The second recv() call overwrites the first three bytes of the buffer with the characters 'r', 'e', 'd', but again, it does not write a string terminator. It follows that if you then print recvline, you get "redlow".

It is important to consider the recv() function's return value. It informs you about several things that you really need to know:

• on a successful reception of one or more bytes, it tells you exactly how many bytes were received. (You can use this in your program to know where to put a string terminator.)

• on an error, including a fairly benign one such as the function being interrupted by a signal before transferring any bytes, it notifies you that an error in fact occurred (by being the value -1; and errno then tells you what kind of error).

• when no more data are available as a result of the remote end having closed the connection, it notifies you of this (by being 0).

It is also important to understand that on a stream-oriented socket such as one based on TCP, send() and recv() are not message-oriented. It is not safe to assume that each send() on one side pairs perfectly with one recv() on the other. Data sent by one send() may be split across more than one recv(), and data sent by multiple send()s may be received by a single recv().

Nor is it in general safe to assume that a send() or recv() will transfer the full number of bytes requested. Both partial sends and partial receives are possible.

Youre pretty close. move totalNum = 0 to be a class level variable, i.e.

Instead of using a function httpserver , I used class and it build the exe without any problem and now the http server runs even in its executable form.

Credit to: https://stackoverflow.com/users/642070/tdelaney for providing this solution at :

https://pastebin.com/KsTmVWRZ

How do i configure docker swarm or another load balancer to somehow send real client's ip to my server?

Docker has already an open issue for that, see https://github.com/docker/roadmap/issues/157 . It is currently not possible to do any configuration to do that.

As you already mentioned earlier, you will have to use some custom way of handling that until the issue is closed.

You can explicitly said you can have any number of call to getAddress: