I have a single Fedora Core 6 macnine with 3 FQDNs on it, primarily I use
it for my ohmster.com play domain. I named my machine "ohmster.com" and
the mail worked. I could get system mail sent to root, aliased to
ohmster, and I could pop3 right into dovecot and retrieve my email via
pop3. root@ohmster.com aliased to ohmster@ohmster.com and I would get the
mail. I have 27/7 cablemodem and use zoneedit for DNS, they also do mail
forwards to my domain to me at my ISP email address. I stopped using
sendmail on the net directly when I got blocked by my previous ISP, the
made packets sent to port 25 "vanish". Since switching to cablemodem, I
can now use port 25 again if I want to but then I get contantly hacked at
to try and use my mail relays. I just use the smarthost feature in
sendmail.cf to bounce mail through my ISP mail host and it works fine.

Recently in the newsreader group, I was told to name my machines on the
net. As such:

> Only if the output of "hostname -f" is a FQDN - which in practice
> means that it should contain two or more dots.

So if I do that I get:
[ohmster@ohmster ~]$ hostname -f
ohmster.com
[ohmster@ohmster ~]$

Said I should change hosts to give my machine a name on the network like
this:
[ohmster@ohmster ~]$ cat /etc/hosts
# Do not remove the following line, or various programs
# that require network functionality will fail.
127.0.0.1 ohmster.com localhost ohmster
192.168.15.100 ohmster.com ohmster
192.168.0.3 missy
192.168.0.2 paula
[ohmster@ohmster ~]$

Change to:

127.0.0.1 ohmster.ohmster.com localhost ohmster
192.168.15.100 ohmster.ohmster.com ohmster
192.168.0.3 missy.ohmster.com missy
192.168.0.2 paula.ohmster.com paula

NOTE: Explain network is in order. I have a cablemodem that goes directly
to the Linux machine on eth0. The Linux machine uses firestarter to give
dhcp to all other machines as well as IP Masquerade out through eth1,
going to a high speed swtich and all other machines. I can plug any
machine into the hub and it will work with all TCP settings on automatic.
I needed Vonage for phone service and had to install their router,
Linksys WRT54GP2, after my modem and before the Linux box. I want the
Linux box to get the "real IP" and put it into the DMZ but it offers my
Linux box an IP of 192.168.15.100. My Linux machine gives out IPs in the
192.168.0.xxx range so my personal Windows machine, missy, is
192.168.0.3.

Since making the changes to the machine's names, I "broke" the mail. See
here where I reply to Blinky about this:

> Oh shoot, that machine renaming did not work at all, Blinky. I get

Bummer.

Undo those changes.

-----------------------------------------------------------------------

Oh shoot, that machine renaming did not work at all, Blinky. I get
system messages from my Linux machine, you know, stuff like Logwatch,
etc., and sendmail never had a problem with it. The mail would go to
root at the machine. Since it was named "ohmster.com", the mail would go
to "root@ohmster.com". That is a valid email address and it would get
sent to root's mailbox and sit there. I would check my mail and dovecot
runs a pop3 port for me to check mail and I get the root email like that
because root is aliased to ohmster.

Now that the machine is named ohmster and is at ohmster.com, the mail
was sent to "root@ohmster.ohmster.com" which is not a valid email
address and my own machine would not take it! It got bounced back to
postmaster at ohmster.com, which had to go out on the Internet to find
from my DNS and finally bounced back to me as ohmster because postmaster
is aliased to me again. This time the mail had to go out in the world to
find out where to bounce it to. Check this out:

++++++++++++++++++++++++++++++++++++++++++++++++++ +++++++++
The original message was received at Sat, 25 Aug 2007 04:52:11 -0400
from localhost
with id l7P8qBh5018194

----- The following addresses had permanent fatal errors -----
<root@ohmster.ohmster.com>
(reason: 550 [PERMFAIL] destination not valid within DNS)

----- Transcript of session follows -----
.... while talking to smtp.comcast.net.:
>>> DATA
<<< 550 [PERMFAIL] destination not valid within DNS
550 5.1.1 <root@ohmster.ohmster.com>... User unknown
<<< 503 need RCPT command [data]
++++++++++++++++++++++++++++++++++++++++++++++++++ +++++++++

So much for nameing the machine "ohmster" or anything else. If I do
that, my own mail won't work.

That sounded so cool too. I really wanted to do that by having my
machine named at my own domain, and have the other machines named at the
same domain, too. Well, only missy would work, paula is offline right
now. But it fucks up the mail so bad that I cannot do it, I am doing
something wrong, might have to go over to comp.mail.sendmail or
something for that kind of if I cannot get it in here.

So we are back to this again:

[ohmster@ohmster etc]$ cat hosts
# Do not remove the following line, or various programs
# that require network functionality will fail.
127.0.0.1 ohmster.com localhost ohmster
192.168.15.100 ohmster.com ohmster
192.168.0.3 missy
192.168.0.2 paula
[ohmster@ohmster etc]$

I did a quick test with:

[ohmster@ohmster etc]$ mail root
Subject: test mail
Testing the email system by sending a test message to root.

Cc:
[ohmster@ohmster etc]$

After putting in a quick one liner "Testing the email system by sending
a test message to root.", I closed the mail message with ctl-d and off
it went. I checked the mail with my regular pop3 mail program, OE, and
the mail was there as normal, no bounces this time. ...sigh.

-----------------------------------------------------------------------

Bottom Line: How should I be naming the machines on my network? Shouldn't
each machine have it's own unique name on the local network such as
paula.ohmster.com, ohmster.ohmster.com, & paula.ohmster.com and if I do
that, How do I get the mail to work and then not have root or user mail
sent to :
"root@ohmster.ohmster.com"?

The other machines use webmail but my machine, missy, uses Outlook
Express to get mail directly from my ISP via pop3 and also check my
ohmster machine with dovecot for pop3 mail.

-----------------------------------------------------------------------
How should I name my machines and set this up, for real, so that it is
considered "proper" and so that it works?

Thanks.
--
~Ohmster * ohmster /a/t/ ohmster dot com
Put "messageforohmster" in message body
(That is Message Body, not Subject!)
to pass my spam filter.

In article <Xns999AC3CC2F9C8MyBigKitty@194.177.96.26> Ohmster
<nowayin@hell.com> writes:
[snip]

>127.0.0.1 ohmster.com localhost ohmster

Not really part of your problem AFAICS, but don't do that - putting
actual host/domain names on the 127.0.0.1 line is some hack that some
Linux distributions have for hosts that don't have any actual network
connection. In any real network setup, the line should only have
'localhost'.

>Bottom Line: How should I be naming the machines on my network? Shouldn't
>each machine have it's own unique name on the local network such as
>paula.ohmster.com, ohmster.ohmster.com, & paula.ohmster.com and if I do
>that, How do I get the mail to work and then not have root or user mail
>sent to :
>"root@ohmster.ohmster.com"?

The short answer is probably that you don't *name* hosts by putting
things in /etc/hosts, that's just a list of names that hosts have been
given and the corresponding IP addresses. If you actually set the
hostname to ohmster.ohmster.com, sendmail should (after a restart)
automatically know to deliver mail for user@ohmster.ohmster.com
locally.

The hostname is set with the (surprise!) 'hostname' command, but of
course you want the system boot scripts to do that based on the contents
of some file, the name of which varies between Unix versions and
probably between Linux distributions - I believe it's
/etc/sysconfig/network on Fedora. Of course you should still have a
matching /etc/hosts (unless you use DNS for local host name/address
resolution) - your attempted changed version seemed OK except for the
127.0.0.1 line.

--Per Hedeland
per@hedeland.org

per@hedeland.org (Per Hedeland) wrote in news:fb5o6j$t7m$1@hedeland.org:

Ahhh, Per Hedeland! It has been a while but how nice to hear from you
again. Yes, you did me very much in the past and for that I am
grateful.

> In article <Xns999AC3CC2F9C8MyBigKitty@194.177.96.26> Ohmster
> <nowayin@hell.com> writes:
> [snip]
>
>>127.0.0.1 ohmster.com localhost ohmster
>
> Not really part of your problem AFAICS, but don't do that - putting
> actual host/domain names on the 127.0.0.1 line is some hack that some
> Linux distributions have for hosts that don't have any actual network
> connection. In any real network setup, the line should only have
> 'localhost'.

So change it back to this?
127.0.0.1 localhost

>>Bottom Line: How should I be naming the machines on my network?
>>Shouldn't each machine have it's own unique name on the local network
>>such as paula.ohmster.com, ohmster.ohmster.com, & paula.ohmster.com
>>and if I do that, How do I get the mail to work and then not have root
>>or user mail sent to :
>>"root@ohmster.ohmster.com"?
>
> The short answer is probably that you don't *name* hosts by putting
> things in /etc/hosts, that's just a list of names that hosts have been
> given and the corresponding IP addresses. If you actually set the
> hostname to ohmster.ohmster.com, sendmail should (after a restart)
> automatically know to deliver mail for user@ohmster.ohmster.com
> locally.

Yeah but that does not make sense. I want the mail to go to
ohmster@ohmster.com, not "ohmster@ohmster.ohmster.com. I own the domain
ohmster.com and want the mail on that machine to be from ohmster.com and
mail to ohmster.com will get intercepted by my free DNS provider,
zoneedit.com. They forward all "ohmster.com" mail to my ISP account and I
get it that way. I used to run sendmail as a real server on the net but
my last ISP blocked port 25, putting an end to my mail server days. Now
that I have a cablemodem, I can do that again if I want but then there is
the constant battle of hackers trying to gain access to the mail server
or use it for a relay.

> The hostname is set with the (surprise!) 'hostname' command, but of
> course you want the system boot scripts to do that based on the
> contents of some file, the name of which varies between Unix versions
> and probably between Linux distributions - I believe it's
> /etc/sysconfig/network on Fedora. Of course you should still have a
> matching /etc/hosts (unless you use DNS for local host name/address
> resolution) - your attempted changed version seemed OK except for the
> 127.0.0.1 line.

Well let's see what is in the /etc/sysconfig/network on this Fedora
machine.

[ohmster@ohmster ~]$ cat /etc/sysconfig/network
NETWORKING_IPV6=no
NETWORKING=yes
GATEWAYDEV=eth0
HOSTNAME=ohmster

[ohmster@ohmster ~]$

Is that not correct? Where does the .com come into the equation? So far,
I am doing that with the /etc/hosts file and my hostname command returns
"ohmster.com".

[ohmster@ohmster ~]$ hostname -f
ohmster.com
[ohmster@ohmster ~]$

The system I am using actually works but is fundamentally not right, I
think. If I had 3 machines on this LAN and they were all nodes of
ohmster.com, then they should all have unique network names, don't you
think? To make the mail work again I have my /etc/hosts file setup as
such:

[ohmster@ohmster ~]$ cat /etc/hosts
# Do not remove the following line, or various programs
# that require network functionality will fail.
127.0.0.1 ohmster.com localhost ohmster
192.168.15.100 ohmster.com ohmster
192.168.0.3 missy
192.168.0.2 paula
[ohmster@ohmster ~]$

Here is a DIA diagram of how my network is setup:
http://www.ohmster.com/~ohmster/picture/OhmNET.jpg

The Fedora computer acts as a router for the home LAN. All home computer
network traffic passes through the Linux box.

So give me your take on this Per Hedeland. How should these machines be
configured to all be a part of ohmster.com such as ohmster.ohmster.com,
paula.ohmster.com, and missy.ohmster.com or is all of this not necessary
because paula and missy are not accessible to the Internet from the
outside because they all have NAT addresses assigned by the firewall and
dchp server on the Fedora box? I want for ohmster.ohmster.com to "be"
ohmster.com and mail sent from there will come from user@ohmster.com and
if I were to setup my mail.ohmster.com DNS and MX record again, to
receive mail to ohmster.com as long as a valid user exists on the system.

If you have some spare time, please give me a rundown Per Hedeland, your
opinions and expertise have gotten my respect and attention a long time
ago. How should /etc/hosts be setup? How should /etc/sysconfig/network be
setup? ohmster.com is a FQDN so I want for this to be done right. As it
is, any mail to or from user@machine.ohmster.com will fail. Is there some
sort of DNS that needs to be setup or just config the machines and
network properly? Can you me?

I am doing this to learn as much as I am to get my system running right.
Thank you my friend.

> --Per Hedeland
> per@hedeland.org
>



--
~Ohmster * ohmster /a/t/ ohmster dot com
Put "messageforohmster" in message body
(That is Message Body, not Subject!)
to pass my spam filter.

On 2007-08-30, Per Hedeland <per@hedeland.org> wrote:
> In article <Xns999AC3CC2F9C8MyBigKitty@194.177.96.26> Ohmster
><nowayin@hell.com> writes:
> [snip]

Per Hedeland, I did as you said and redid my /etc/hosts file.

>>127.0.0.1 ohmster.com localhost ohmster
>
> Not really part of your problem AFAICS, but don't do that - putting
> actual host/domain names on the 127.0.0.1 line is some hack that some
> Linux distributions have for hosts that don't have any actual network
> connection. In any real network setup, the line should only have
> 'localhost'.

This is what I have now:

[ohmster@ohmster ~]$ cat /etc/hosts
# Do not remove the following line, or various programs
# that require network functionality will fail.
127.0.0.1 localhost
192.168.15.100 ohmster.ohmster.com ohmster #eth0 to Internet
192.168.0.1 ohmster.ohmster.com ohmster #eth1 to LAN
192.168.0.3 missy
192.168.0.2 paula
[ohmster@ohmster ~]$ hostname -f
ohmster.ohmster.com
[ohmster@ohmster ~]$


>>Bottom Line: How should I be naming the machines on my network? Shouldn't
>>each machine have it's own unique name on the local network such as
>>paula.ohmster.com, ohmster.ohmster.com, & paula.ohmster.com and if I do
>>that, How do I get the mail to work and then not have root or user mail
>>sent to :
>>"root@ohmster.ohmster.com"?
>
> The short answer is probably that you don't *name* hosts by putting
> things in /etc/hosts, that's just a list of names that hosts have been
> given and the corresponding IP addresses. If you actually set the
> hostname to ohmster.ohmster.com, sendmail should (after a restart)
> automatically know to deliver mail for user@ohmster.ohmster.com
> locally.

I did restart sendmail this time, as well as the network, and the
firewall. Sendmail did get the mail right this time.

> The hostname is set with the (surprise!) 'hostname' command, but of
> course you want the system boot scripts to do that based on the contents
> of some file, the name of which varies between Unix versions and
> probably between Linux distributions - I believe it's
> /etc/sysconfig/network on Fedora. Of course you should still have a
> matching /etc/hosts (unless you use DNS for local host name/address
> resolution) - your attempted changed version seemed OK except for the
> 127.0.0.1 line.
>
> --Per Hedeland
> per@hedeland.org

[ohmster@ohmster ~]$ cat /etc/sysconfig/network
NETWORKING_IPV6=no
NETWORKING=yes
GATEWAYDEV=eth0
HOSTNAME=ohmster.ohmster.com

[ohmster@ohmster ~]$

missy is my XP machine on the LAN, paula is not online right now. These
Windows machines are simply named as "MISSY" and "PAULA" as per "My
Computer/Properties/Computer Name Tab". As far as Windows goes, they are
not named "missy.ohmster.com.

My firewall bitches about this (I think) when it restarts. See firewall
(Firestarter) output:

[root@ohmster etc]# cat dhclient-exit-hooks
# sh /etc/firestarter/firestarter.sh stop
# sh /etc/firestarter/firestarter.sh start
# /home/ohmster/scripts/domain_IP_update
[root@ohmster etc]# sh /etc/firestarter/firestarter.sh stop
Firewall stopped
[root@ohmster etc]# sh /etc/firestarter/firestarter.sh start
iptables v1.3.5: host/network ` missy' not found
Try `iptables -h' or 'iptables --' for more information.
iptables v1.3.5: host/network ` missy' not found
Try `iptables -h' or 'iptables --' for more information.
iptables v1.3.5: Bad IP address ` missy'

Try `iptables -h' or 'iptables --' for more information.
iptables v1.3.5: Bad IP address ` missy'

Try `iptables -h' or 'iptables --' for more information.
Firewall started
[root@ohmster etc]#

See? "missy" not found. Everything works so far though on the network
and so far as the mail goes.

Should these Windows machines be renamed as to be part of the domain?
When I try to do this by using the Network Wizard on Windows, I am given
the choice that this is a business network computer and I use it to
connect to other computers at work. Then My company uses a network with
a domain (ohmster.com).

Then I am told that to connect to a Windows network I need all kinds of
stuff like User name, password, user account domain, computer name, and
computer domain. What is all this stuff and is this necessary to join a
Linux domain network?

Is all of this necessary to configure my network properly so far that
the mail works and the network is done correctly?

Thanks anybody for your , this is a bit over my head and I want to
get it right.
--
~Ohmster | ohmster /a/t ohmster dot com
Put "messageforohmster" in message body
(That is MESSAGE BODY, not Subject!)
to pass my spam filter.

On Sat, 01 Sep 2007 07:38:10 +0200, Ohmster wrote:
> This is what I have now:
>
> [ohmster@ohmster ~]$ cat /etc/hosts
> # Do not remove the following line, or various programs
> # that require network functionality will fail.
> 127.0.0.1 localhost
> 192.168.15.100 ohmster.ohmster.com ohmster #eth0 to Internet
> 192.168.0.1 ohmster.ohmster.com ohmster #eth1 to LAN
> 192.168.0.3 missy
> 192.168.0.2 paula
> [ohmster@ohmster ~]$ hostname -f
> ohmster.ohmster.com
> [ohmster@ohmster ~]$
[snip]

Every interface must have an unique name. You have 2 the same. Fix that
first...

Ohmster wrote:

> My firewall bitches about this (I think) when it restarts. See firewall
> (Firestarter) output:
>
> [root@ohmster etc]# cat dhclient-exit-hooks
> # sh /etc/firestarter/firestarter.sh stop
> # sh /etc/firestarter/firestarter.sh start
> # /home/ohmster/scripts/domain_IP_update
> [root@ohmster etc]# sh /etc/firestarter/firestarter.sh stop
> Firewall stopped
> [root@ohmster etc]# sh /etc/firestarter/firestarter.sh start
> iptables v1.3.5: host/network ` missy' not found
> Try `iptables -h' or 'iptables --' for more information.
> iptables v1.3.5: host/network ` missy' not found
> Try `iptables -h' or 'iptables --' for more information.
> iptables v1.3.5: Bad IP address ` missy'
>
> Try `iptables -h' or 'iptables --' for more information.
> iptables v1.3.5: Bad IP address ` missy'
>
> Try `iptables -h' or 'iptables --' for more information.
> Firewall started
> [root@ohmster etc]#
>
> See? "missy" not found.

No, the errer message is: ` missy' not found
Note the space between the quote signs in front
of the name.
Look in your firewall rules for occurrences
of: " missy"
or: ' missy'
or even: \ missy


Regards,

Kees.

--
Kees Theunissen.

"Steve" <steve@yobank.com> wrote in message news:fbavg9$i3i$3@lust.ihug.co.nz...
> On Sat, 01 Sep 2007 07:38:10 +0200, Ohmster wrote:
> > This is what I have now:
> >
> > [ohmster@ohmster ~]$ cat /etc/hosts
> > # Do not remove the following line, or various programs
> > # that require network functionality will fail.
> > 127.0.0.1 localhost
> > 192.168.15.100 ohmster.ohmster.com ohmster #eth0 to Internet
> > 192.168.0.1 ohmster.ohmster.com ohmster #eth1 to LAN
> > 192.168.0.3 missy
> > 192.168.0.2 paula
> > [ohmster@ohmster ~]$ hostname -f
> > ohmster.ohmster.com
> > [ohmster@ohmster ~]$
> [snip]
>
> Every interface must have an unique name. You have 2 the same. Fix that
> first...

No, that should be OK insofar as anything which looks up by IP address
192.168.15.100 or 192.168.0.1 will get ohmster, and anything querying
by name ohmster will get 192.168.15.100.

The question is whether routing is set up appropriately so that everything
for 192.168.0 goes to the switch and everything else goes via the default
gateway which is (or should be) the router, neither of which is in /etc/hosts.
Depending on netmasks and switch and router addresses, that might require
a static route. Also, the OP mentioned NAT-ing or IP address masquerading
through eth1 which presumably ought to be eth0.

Alternatively, the switch could be plugged straight into the router so
it is not necessary to route all traffic through the linux server. This would
mean relying on the router's firewall, which the OP is perhaps not
prepared to do but it would simplify things a great deal (oh, and probably
the switch and certainly the router should be able to act as dhcp server).

The OP also mentioned a DNS lookup being necessary to find one
of the machines which implies nsswitch.conf is wrongly not configured
to look at hosts, though that might have been before hosts was set up.

On the Windows servers, you probably need to specify tcp/ip
networking in order to set things up appropriately.

These network problems are probably off-topic for comp.mail.sendmail.

The mail setup seemed slightly odd with local mail going out and
coming back in again but I have probably misunderstood the OP.
Masquerading (sendmail not ip) can be set up so that mail from
user@ohmster.ohmster.com comes from user@ohmster.com.
The OP should note that root is normally excluded from masquerading
as you generally do want to know which machine's root is sending
mail to alert you to a system problem (imagine buying a second
linux server).

On the question of outgoing mail aliases, just set them to whatever
is wanted but remember to run newaliases afterwards. It looked like
the OP had aliased to the local user called user (bad example name!)
but wanted to alias to user@ohmster.com

Or it may be that the OP wants no local mail deliveries at all,
with all mail forwarded to the ISP, a bit like
http://www.harker.com/sendmail/submit.html
Is the ISP or the linux server supposed to be the mail server?

--
John.

"John L" <jl@lammtarra.notthisbit.fslife.co.uk> wrote in news:46d92b76$0
$648$bed64819@news.gradwell.net:

> The mail setup seemed slightly odd with local mail going out and
> coming back in again but I have probably misunderstood the OP.
> Masquerading (sendmail not ip) can be set up so that mail from
> user@ohmster.ohmster.com comes from user@ohmster.com.
> The OP should note that root is normally excluded from masquerading
> as you generally do want to know which machine's root is sending
> mail to alert you to a system problem (imagine buying a second
> linux server).

sendmail is used to send mail on this server. I receive mail on my ISP
server because my DNS service does "mail forward" in that any mail sent to
my domain, gets redirected to my ISP's pop3 mail server.


> On the question of outgoing mail aliases, just set them to whatever
> is wanted but remember to run newaliases afterwards. It looked like
> the OP had aliased to the local user called user (bad example name!)
> but wanted to alias to user@ohmster.com

That was a "made up" example. I did not use that example actually.

> Or it may be that the OP wants no local mail deliveries at all,
> with all mail forwarded to the ISP, a bit like
> http://www.harker.com/sendmail/submit.html
> Is the ISP or the linux server supposed to be the mail server?
>

The ISP receives mail through mail redirect from the DNS server
(zoneedit.com) and sendmail will send mail out directly and uses my ISP
host as a smart_host.

--
~Ohmster * ohmster /a/t/ ohmster dot com
Put "messageforohmster" in message body
(That is Message Body, not Subject!)
to pass my spam filter.
..

Kees Theunissen <theuniss@rijnh.nl> wrote in news:46d91796$0$232
$e4fe514c@news.xs4all.nl:

>> My firewall bitches about this (I think) when it restarts. See
firewall
>> (Firestarter) output:
>>
>> [root@ohmster etc]# cat dhclient-exit-hooks
>> # sh /etc/firestarter/firestarter.sh stop
>> # sh /etc/firestarter/firestarter.sh start
>> # /home/ohmster/scripts/domain_IP_update
>> [root@ohmster etc]# sh /etc/firestarter/firestarter.sh stop
>> Firewall stopped
>> [root@ohmster etc]# sh /etc/firestarter/firestarter.sh start
>> iptables v1.3.5: host/network ` missy' not found
>> Try `iptables -h' or 'iptables --' for more information.
>> iptables v1.3.5: host/network ` missy' not found
>> Try `iptables -h' or 'iptables --' for more information.
>> iptables v1.3.5: Bad IP address ` missy'
>>
>> Try `iptables -h' or 'iptables --' for more information.
>> iptables v1.3.5: Bad IP address ` missy'
>>
>> Try `iptables -h' or 'iptables --' for more information.
>> Firewall started
>> [root@ohmster etc]#
>>
>> See? "missy" not found.
>
> No, the errer message is: ` missy' not found
> Note the space between the quote signs in front
> of the name.
> Look in your firewall rules for occurrences
> of: " missy"
> or: ' missy'
> or even: \ missy
>
>
> Regards,
>
> Kees.

I have been all through my firewall rules and cannot find ' missy'
anywhere with a space in front of it.

[root@ohmster inbound]# iptables -L |grep "missy"
ACCEPT tcp -- anywhere missy tcp dptms
ACCEPT udp -- anywhere missy udp dptms
ACCEPT tcp -- anywhere missy tcp dpt:rfa
ACCEPT udp -- anywhere missy udp dpt:rfa
ACCEPT tcp -- anywhere missy tcp dpt:nsws
ACCEPT udp -- anywhere missy udp dpt:nsws
ACCEPT tcp -- anywhere missy tcp dpt:4552
ACCEPT udp -- anywhere missy udp dpt:4552
ACCEPT tcp -- anywhere missy tcp dpt:llm-
pass
ACCEPT udp -- anywhere missy udp dpt:llm-
pass
ACCEPT all -- missy anywhere
ACCEPT tcp -- missy anywhere tcp dpt:ndmp
ACCEPT udp -- missy anywhere udp dpt:ndmp
ACCEPT tcp -- missy anywhere tcp dpt:snmp
ACCEPT udp -- missy anywhere udp dpt:snmp
[root@ohmster inbound]# iptables -L |grep " missy"
ACCEPT tcp -- anywhere missy tcp dptms
ACCEPT udp -- anywhere missy udp dptms
ACCEPT tcp -- anywhere missy tcp dpt:rfa
ACCEPT udp -- anywhere missy udp dpt:rfa
ACCEPT tcp -- anywhere missy tcp dpt:nsws
ACCEPT udp -- anywhere missy udp dpt:nsws
ACCEPT tcp -- anywhere missy tcp dpt:4552
ACCEPT udp -- anywhere missy udp dpt:4552
ACCEPT tcp -- anywhere missy tcp dpt:llm-
pass
ACCEPT udp -- anywhere missy udp dpt:llm-
pass
ACCEPT all -- missy anywhere
ACCEPT tcp -- missy anywhere tcp dpt:ndmp
ACCEPT udp -- missy anywhere udp dpt:ndmp
ACCEPT tcp -- missy anywhere tcp dpt:snmp
ACCEPT udp -- missy anywhere udp dpt:snmp
[root@ohmster inbound]#


--
~Ohmster * ohmster /a/t/ ohmster dot com
Put "messageforohmster" in message body
(That is Message Body, not Subject!)
to pass my spam filter.

"John L" <jl@lammtarra.notthisbit.fslife.co.uk> wrote in
news:46d92b76$0$648$bed64819@news.gradwell.net:

>
> "Steve" <steve@yobank.com> wrote in message
> news:fbavg9$i3i$3@lust.ihug.co.nz...
>> On Sat, 01 Sep 2007 07:38:10 +0200, Ohmster wrote:
>> > This is what I have now:
>> >
>> > [ohmster@ohmster ~]$ cat /etc/hosts
>> > # Do not remove the following line, or various programs
>> > # that require network functionality will fail.
>> > 127.0.0.1 localhost
>> > 192.168.15.100 ohmster.ohmster.com ohmster #eth0 to Internet
>> > 192.168.0.1 ohmster.ohmster.com ohmster #eth1 to LAN
>> > 192.168.0.3 missy
>> > 192.168.0.2 paula
>> > [ohmster@ohmster ~]$ hostname -f
>> > ohmster.ohmster.com
>> > [ohmster@ohmster ~]$
>> [snip]
>>
>> Every interface must have an unique name. You have 2 the same. Fix
>> that first...
>
> No, that should be OK insofar as anything which looks up by IP address
> 192.168.15.100 or 192.168.0.1 will get ohmster, and anything querying
> by name ohmster will get 192.168.15.100.
>
> The question is whether routing is set up appropriately so that
> everything for 192.168.0 goes to the switch and everything else goes
> via the default gateway which is (or should be) the router, neither of
> which is in /etc/hosts. Depending on netmasks and switch and router
> addresses, that might require a static route. Also, the OP mentioned
> NAT-ing or IP address masquerading through eth1 which presumably ought
> to be eth0.

I used DIA to make a diagram of my network setup. You can view it here:
http://www.ohmster.com/~ohmster/picture/OhmNET.jpg

> Alternatively, the switch could be plugged straight into the router so
> it is not necessary to route all traffic through the linux server.
> This would mean relying on the router's firewall, which the OP is
> perhaps not prepared to do but it would simplify things a great deal
> (oh, and probably the switch and certainly the router should be able
> to act as dhcp server).

I want to learn Linux routing and networking so this is the reason that I
make the Linux box the router for the network. I really do not need this
as the Linksys can do this for me, but I use the Linksys to get Vonage
phone to work. I really want for the Linksys to pass the real IP through
to the Linux box by putting it in the DMZ but so far, it does not work. I
can only get an IP of 192.168.15.100 from the Linksys for the Linux box.

> The OP also mentioned a DNS lookup being necessary to find one
> of the machines which implies nsswitch.conf is wrongly not configured
> to look at hosts, though that might have been before hosts was set up.

I am not sure what that means.

> On the Windows servers, you probably need to specify tcp/ip
> networking in order to set things up appropriately.

The Windows boxes work fine and they do use tcp/ip for networking. I was
thinking of making them join the Linux domain but the how-tos that I have
found seem to indicate that you can make them join a samba domain. Right
now, I have the Windows boxes work with samba by specifying the
workgroup. These tutorials explain how to do it as a domain. I tried it
and it did not find a domain server on the Linux box. I must have samba
setup incorrectly for this, not sure how to fix it.

http://rudd-o.com/archives/2006/03/0...-join-a-samba-
domain-in-5-minutes/
or
http://tinyurl.com/yns7o3

And

http://www.samba.org/samba/docs/man/Samba-HOWTO-
Collection/ClientConfig.html#id2545087
or
http://tinyurl.com/37v6gf

> These network problems are probably off-topic for comp.mail.sendmail.

Probably, I might have to go to the networking newsgroup for this.

> The mail setup seemed slightly odd with local mail going out and
> coming back in again but I have probably misunderstood the OP.
> Masquerading (sendmail not ip) can be set up so that mail from
> user@ohmster.ohmster.com comes from user@ohmster.com.
> The OP should note that root is normally excluded from masquerading
> as you generally do want to know which machine's root is sending
> mail to alert you to a system problem (imagine buying a second
> linux server).

I used to have sendmail and mx records setup so that I had a real mail
server but my ISP at that time killed all traffic on port 25 to put an
end to spammers getting cheap DSL accounts and then flooding the net with
spam. My new ISP has no such restriction and I can do that again but then
I have to deal with spammers trying to crack my mail server to relay
their crap spam mail. I use smart_host with sendmail to send mail out
through my ISP and my DNS service (zoneedit.com) does mail forward to
send all my domain mail to me via my ISP's mail server. It works pretty
good that way.

> On the question of outgoing mail aliases, just set them to whatever
> is wanted but remember to run newaliases afterwards. It looked like
> the OP had aliased to the local user called user (bad example name!)
> but wanted to alias to user@ohmster.com

It was just an example. I did not do this for real. Let me send mail to
myself and see how it comes...

Received: from ohmster.ohmster.com (c-71-57-187-110.hsd1.fl.comcast.net
[71.57.187.110])
by comcast.net (alnrmhc14) with ESMTP
id <20070901235857b1400ph4h5e>; Sat, 1 Sep 2007 23:58:57 +0000
Received: from ohmster.ohmster.com (localhost [127.0.0.1])
by ohmster.ohmster.com (8.13.8/8.13.8) with ESMTP id l81NwuuN026966
for <theohmster@comcast.net>; Sat, 1 Sep 2007 19:58:56 -0400
Received: from localhost (ohmster@localhost)
by ohmster.ohmster.com (8.13.8/8.13.8/Submit) with ESMTP id
l81NwtPY026963
for <theohmster@comcast.net>; Sat, 1 Sep 2007 19:58:56 -0400
X-Authentication-Warning: ohmster.ohmster.com: ohmster owned process
doing -bs
Date: Sat, 1 Sep 2007 19:58:55 -0400 (EDT)
From: Ohmster <ohmster@ohmster.com>
To: Ohmster <theohmster@comcast.net>
Subject: Test Mail

Yeah, that is working pretty much the way I want.

> Or it may be that the OP wants no local mail deliveries at all,
> with all mail forwarded to the ISP, a bit like
> http://www.harker.com/sendmail/submit.html
> Is the ISP or the linux server supposed to be the mail server?
>

ISP for incoming mail (Routed by DNS service as a mail forward,
zoneedit.com) and Linux server to send mail directly via smart_host. I am
doing this to keep the rottem spammers from tring to crack my mail server
and then use it as a relay.

Thanks for your .
--
~Ohmster * ohmster /a/t/ ohmster dot com
Put "messageforohmster" in message body
(That is Message Body, not Subject!)
to pass my spam filter.