Question

I am writing a c++ program that will store everything in a mysql database. I would like to avoid the Windows Security window from popping up and asking if mysqld can have priviledges. If I click cancel (in response to the firewall window), the program still runs just fine because everything is local.

I am trying to use named pipes with the –skip-networking flag to avoid this firewall window.

START EDIT:

I am using a portable version of mysqld. It won’t have admin priviledges. (Thanks to Ben for pointing out that this is an important piece of information)

END EDIT

I am using the following command to start the mysqld server:

C:\tmp_report\12345\mysqld\mysql\bin\mysqld --user=root \
--basedir="C:\tmp_report\12345\mysqld\\mysql" \
--datadir="C:\tmp_report\12345\mysqld\data" \
--pid-file="C:\tmp_report\12345\mysqld\mysql.pid" \
--log-error="C:\tmp_report\12345\mysqld\mysql.err" \
--skip-networking \
--enable-named-pipe

This works and I am able to connect to the database using the following command line statement:

mysql --pipe -uroot 

I am trying to connect through the c-api as follows:

unsigned int mysql_timeout=200;
unsigned int protocol=MYSQL_PROTOCOL_PIPE;
bool mysql_reconnect=true;
bool mysql_local_infile=true;

if (mysql_options(conn,MYSQL_OPT_CONNECT_TIMEOUT, (char *)&mysql_timeout)){*logstream<<"mysql option error connect_timeout";}

if (mysql_options(conn,MYSQL_OPT_READ_TIMEOUT, (char *)&mysql_timeout)){*logstream<<"mysql option error read_timeout";}

if (mysql_options(conn,MYSQL_OPT_WRITE_TIMEOUT, (char *)&mysql_timeout)){*logstream<<"mysql option error write_timeout";}

if (mysql_options(conn,MYSQL_OPT_RECONNECT,                   &mysql_reconnect)){*logstream<<"mysql option error reconnect";}

if (mysql_options(conn,MYSQL_OPT_LOCAL_INFILE,                &mysql_local_infile)){*logstream <<"mysql option error local_infile";}

if (mysql_options(conn,MYSQL_OPT_NAMED_PIPE,                  NULL)){*logstream<<"mysql option error MYSQL_OPT_NAMED_PIPE";}
if (mysql_options(conn,MYSQL_OPT_PROTOCOL,            (char *)&protocol))          {*logstream<<"mysql option error protocol";}

if (!mysql_real_connect(conn,
        "localhost", // SERVER
        "root",      // USER
        NULL,   // PASS
        NULL,   // DATABASE
        0,    // port
        NULL, // socket
        0     // client_flag
    )){
    *logstream<<"Database::connect mysql_real_connect failed"<<endl;
    *logstream<<mysql_error(conn)<<endl;
    conn=NULL;
}

The result of this mysql_real_connect call is:

Database::connect mysql_real_connect failed
Lost connection to MySQL server at 'waiting for initial communication packet', system error: 0

If I use the TCPIP connection, everything works fine except that the Windows firewall pops up. I would really like to make this work without requiring the user to click cancel on the firewall window.

Any help is greatly appriciated!

Answer 1

If MySQL is intentionally local only, you should simply configure it to listen only on the local address. Put this in your my.cnf file:

bind-address     = 127.0.0.1

Otherwise use netsh advfirewall or similar to configure the firewall to allow MySQL. This would generally be done when you install MySQL.

Examples:

• http://support.microsoft.com/kb/947709

Or from script:

• http://technet.microsoft.com/en-us/library/cc737845(v=WS.10).aspx

For a single-use installation which doesn’t have or require admin privileges, you can configure local communication using the nonroutable IP address and a random port.

You can achieve this by adding --bind-address=127.0.0.1 --port=62190 to the command line.

You won’t be able to alter the firewall rules unless you are an administrator, and listening on named pipes generally requires administrative privileges too. Creating a shared memory section requires SeCreateGlobalPrivilege, so essentially the same restriction applies.