mettle_get_fqdn: Returns IP address not hostname

[bcoles]

diff --git a/mettle/src/mettle.c b/mettle/src/mettle.c
index ae50704..6bdf22c 100644
--- a/mettle/src/mettle.c
+++ b/mettle/src/mettle.c
@@ -255,6 +255,7 @@ struct mettle *mettle(void)
        m->em = extmgr_new();
 
        sigar_fqdn_get(m->sigar, m->fqdn, sizeof(m->fqdn));
+       printf("FQDN: %s\n", m->fqdn);
 
        sigar_sys_info_get(m->sigar, &m->sysinfo);

$ ./build/linux.x86_64/mettle/src/mettle 
FQDN: 192.168.200.204

[usiegl00]

What is the hostname, this is "expected" behavior if it is not a FQDN:
https://github.com/rapid7/sigar/blob/08507a8aae1d37d040f888fe7a42c166a8732122/src/sigar.c#L2374

[bcoles]

What is the hostname, this is "expected" behavior if it is not a FQDN: https://github.com/rapid7/sigar/blob/08507a8aae1d37d040f888fe7a42c166a8732122/src/sigar.c#L2374

Makes sense. The tested system hostname is not a FQDN.

On the Metasploit side we expect only a hostname. It doesn't need to be a FQDN (rapid7/metasploit-framework#17289). An IP address instead rather than a hostname is unexpected.

[usiegl00]

Maybe break sigar_hostname_get out from sigar_fqdn_get?

[usiegl00]

There is also sigar_net_info_get which has the host_name field.

[sempervictus]

There is also sigar_net_info_get which has the host_name field.

This strikes me as the right direction since it doesnt mess w/ upstream and gives us a few other fields we can use (or discard) as we please.