I have a huge problem with deffered e-mails and I tried smart_host but didn't solve it.

My configuration is the following .
I have a mail server mail.example.gr which is configured with a virtual ip address .227 on eth0:1
I have as well my dsl connection through eth0 .237
All my outgoing e-mail goes out from eth0 (.237) so when a destination mx requests reverse lookup of the ip it end's up in resolving to .237 and the e-mail is deffered.
What I want is all my outgoing e-mail to be routed through my virtual ip which is .227 in eth0:1
When I used
define(`SMART_HOST',`mail.orco.gr')dnl which actually resolves to .227
again my e-mail gets deffered. I know that I am supposed to have my ISP's address here but I thought I'll give it a try.
Ofcourse I have an entry in my /etc/host file.
When I run sendmail -q -v I get the following
unning /var/spool/mqueue/mBN8MtVK031882 (sequence 1 of 1)
<mgennatou@bankofgreece.gr>... Connecting to mail.bankofgreece.gr. via esmtp...
220 mail.bankofgreece.gr ESMTP ready at Tue, 23 Dec 2008 10:24:09 +0200
>>> EHLO mail.orco.gr
250-mail.bankofgreece.gr
250-SIZE 10240000
250-8BITMIME
250 EHLO mail.orco.gr
>>> MAIL From:<npal@orco.gr> SIZE=787
421 mail.bankofgreece.gr has refused your connection because your server did not have a PTR record.
>>> QUIT
<mgennatou@bankofgreece.gr>... Connecting to mail-relay.bankofgreece.gr. via esmtp...
<mgennatou@bankofgreece.gr>... Deferred: Connection timed out with mail-relay.bankofgreece.gr.
Closing connection to mail.bankofgreece.gr.

Any suggestions would be more than helpfull because I am getting crazy. I am trying to resolve this for about ten days.

It seems that both your IPs (x.x.x.227 and x.x.x.237) do not have a PTR record. To be more precise .227 can be resolved only from your ISP's dns and not from other dns servers.
You should contact your ISP to tell them about this problem and maybe ask them to allow you to do the reverse dns for your subnet (classless IN-ADDR.ARPA delegation).

Regards

My ISP all ready has a PTR record for the .227 (mail.orco.gr)address and they can't add another PTR record for the .237 addres beucase I'll have a conflict.
From my side I have configured the Client options in sumbit.cf in order to make sure that all my e-mails are leaving with the .227 as their source address. But the MX server at the destination still tries to resolve the .237 address? Which is the IP address of the actual eth0 interface.
Could it be due to routing rules in the cisco router?
I mean how is it possible an e-mail that leaves my server with a header .227 to change during it's trip to the destination.

What conflict? If a single IP address resolves in a fqdn there will be no conflict.
You also can try to use the ClientPortOptions in sendmail.cf to see if it helps:

I have all ready used the ClientPortOptions in sendmail.cf but still the mail gets deferred.
Now what I mean with conflict is that my ISP cannot have a record for
mail.orco.gr pointing to IP xxx.xxx.xxx.227 and another one
mail.orco.gr pointing to IP xxx.xxx.xxx.237.
Do you think it's a routing problem? That's my only guess since the e-mail is leaving with IP .227.
I was thinking to request for my provider to relay e-mails through him but probably that's going to be some extra cost or as you suggested ask them to allow to do the reverse dns for my subnet (classless IN-ADDR.ARPA delegation) which probably will be costless as well.

Try the following varioation to see if it works:Well, he can but it's useless here (this is used for a round robin setup).
What you need is that the IP resolves to a valid hostname. You can have xx.xx.xx.227 PTR pointing to mail.orco.gr and xx.xx.xx.237 PTR pointing again to mail.orco.gr. Or you can use a different name if you want(xx.xx.xx.237 PTR mail2.orco.gr).

As for the rest of your questions:
1. I don't think it's a routing problem.
2. You can use your ISP's mailserver (mailgate.forthnet.gr) as a smarthost without any extra cost.
3. I don't know the kind of connection you use, but I was working in a company using a leased line provided by your ISP and our /27 subnet delegation was free. You better contact them for details.

I decide to go with adding two more PTR records mail2.orco.gr mail3.orco.gr that will resolve to xx.xx.xx.237 and xx.xx.xx.238 because in order to relay my e-mail through forthnet they have specific service which of course they charge.
Is there anything I need to do in my side in order to accommodate these changes.
I mean, does sendmail.cf or mc needs any changes?
By the way thank you for your help so far!!!

It seems that you are not authoritative for the domain in question, but forthnet. Check the output of:Thus any changes you make to your local dns are not reflected to the rest of the world.
You have to either tell forthnet to update their DNSes with your records, or make your dns authoritative (you should do this this through the domain registrar).

Regards

As far as necessary MX and PTR records is concerned, forthnet is going to update their dns records.
What I am asking is if I'll have to change anything on my side. On the configuration of sendmail.
For example I have configured the ClientOptions parameter in my sendmail.cf with the xx.xx.xx.227.
So do I need to reconfigure anything in sendmail?

I don't think that you have to reconfigure anything in sendmail. Of course you'll see if you have any problems now that your A and PRT records are set.
Btw ClientOptions is not necessary any more.

Regards

Thank you very much for all your help. Issue solved !
I didn't need to change anything my side. Everything works now.
Only question now is why this thing happened in the first place.
I mean I did upgrade my distro and my kernel beginning of october (I do upgrades once a year) but I kept the configuration of sendmail exactly the same. Any way I'll solve the mystery.