I am writing an application that is requiring me to do a DNS lookup for an MX record. I'm not sure if anyone has had experience doing this kind of work but if you do, any help would be appreciated.
EDIT: The thing that I'm going for is an application that will send an e-mail alert. The problem is I need to have the application be able to lookup the MX record for a domain.
The simplest method is to simply use commonly available tools.
The basic "dig" command will return the records to you via this query:
dig mx example.com
If you want just the lines with the mx records...
dig mx example.com | grep -v '^;' | grep example.com
dig is available on most linux / unix boxes.
If you're on windows you can use nslookup
nslookup -type=mx example.com
Then just parse the output of these common tools.
Since you put "C" as a tag, I guess you're looking for source code to do MX lookups using raw sockets. I copied this from http://www.developerweb.net/forum/showthread.php?t=3550. It may be more what you're looking for?
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <netinet/in.h>
#include <resolv.h>
int main (int argc, char *argv[])
{
u_char nsbuf[4096];
char dispbuf[4096];
ns_msg msg;
ns_rr rr;
int i, j, l;
if (argc < 2) {
printf ("Usage: %s <domain>[...]\n", argv[0]);
exit (1);
}
for (i = 1; i < argc; i++) {
l = res_query (argv[i], ns_c_any, ns_t_mx, nsbuf, sizeof (nsbuf));
if (l < 0) {
perror (argv[i]);
} else {
#ifdef USE_PQUERY
/* this will give lots of detailed info on the request and reply */
res_pquery (&_res, nsbuf, l, stdout);
#else
/* just grab the MX answer info */
ns_initparse (nsbuf, l, &msg);
printf ("%s :\n", argv[i]);
l = ns_msg_count (msg, ns_s_an);
for (j = 0; j < l; j++) {
ns_parserr (&msg, ns_s_an, j, &rr);
ns_sprintrr (&msg, &rr, NULL, NULL, dispbuf, sizeof (dispbuf));
printf ("%s\n", dispbuf);
}
#endif
}
}
exit (0);
}
I notice that you're writing for Linux. The idomatic way for a regular program to send mail on Unix-like systems is either:
Both ways presume that the local mailer is configured to pass mail on to where it has to go; on a well-configured Linux box this is a fair assumption.
If that doesn't appeal, the second-best way is for your program to accept the address of a local mail relay server to use, and just connect to that server on port 25.
In other words, wherever possible, use an existing mail relay to send your mail on. Those mail relays will have all the local knowledge that might be necessary to get mail out of the network that you're running on - just looking up the MX and trying to send directly to the destination is not always going to work.
If you've read all that and you still want to look up MX records, try the adns library, it takes care of all the tedious details involved in DNS resolution (and believe me, it is tedious, and easy to get wrong!).
[Disclaimer: I used to be a happy beta tester of SnertSoft's stuff, and I'm still running several of their products]
Actually doing this by hand has its pitfalls, e.g. when you have to handle a truncated response and need to switch from UDP to TCP.
SnertSoft's libsnert (free, click-through-license required) already has a C implementation for this returning a vector of entries. Try to avoid "not-invented-here" if possible :)
Do you indeed have to reimplement an MTA instead of using some existing solution?
Take a look at ldns, from NLnet Labs. This library handles low level packet handling and has its own resolver client built-in.
The documentation includes example code that does exactly what you require.
