What is the use of SMTP server
Sony Antony
When I added an email to the .forward file, sendmail faithfully sent
it to the destination.
But then I realized that nowhere had I specified my SMTP server.
So I started investigating using strace ing sendmail.
I found out that sendmail did not use an SMTP server at all. It did an
MX query, got the mail gateway for the destination address domain, and
sent the mail directly.
This makes me think why ISPs give an SMTP server and ask you to
configure your mail client to point to that.
Why is that none of the mail clients like evolution, are built with
the kind of logic as sendmail. Why is that users are asked to do one
more configuration step ( configuring SNTP server for the mail client
) that can be eliminated.
Any pointers will be appreciated.
--sony
Peter T. Breuer
> I found out that sendmail did not use an SMTP server at all. It did an
Only because you didn't configure a smarthost.
> MX query, got the mail gateway for the destination address domain, and
> sent the mail directly.
> This makes me think why ISPs give an SMTP server and ask you to
> configure your mail client to point to that.
Because they generally block outgoing packets aimed at port 25 on other
machines, in order to stop people like you spamming the world. Ask them
why they aren't doing it, if they aren't. If they aren't, they will
likely host a spammer in short order, and have their IP range blocked
for email by the rest of the world as a result, which YOU won'tlike.
> Why is that none of the mail clients like evolution, are built with
> the kind of logic as sendmail.
(a) because tehy are, (b) because sendmail is not a client.
> Why is that users are asked to do one
> more configuration step ( configuring SNTP server for the mail client
> ) that can be eliminated.
(c) because it can't. Stop making wrong assumptions and you will be les
annoying.
Peter
ynotssor
news:3eb007f1.03071...@posting.google.com
> I was playing with sendmail and .forward file on my RH Linux system.
> When I added an email to the .forward file, sendmail faithfully sent
> it to the destination.
> But then I realized that nowhere had I specified my SMTP server.
> So I started investigating using strace ing sendmail.
> I found out that sendmail did not use an SMTP server at all. It did an
> MX query, got the mail gateway for the destination address domain, and
> sent the mail directly.
What are you talking about? sendmail *is* the smtp server on localhost:25 for whatever email client you used, acting as an smtp
client for the destination server.
If you don't specify an smtp server in your client configuration, the default is localhost. The mail transport agent (MTA) can be
sendmail, postfix, exim or whatever you have listening on port 25.
tony
--
use hotmail com for any email replies
-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 80,000 Newsgroups - 16 Different Servers! =-----
Paul Black
> I found out that sendmail did not use an SMTP server at all. It did an
> MX query, got the mail gateway for the destination address domain, and
> sent the mail directly.
As it should, sendmail being an SMTP server.
> This makes me think why ISPs give an SMTP server and ask you to
> configure your mail client to point to that.
An ISP needs an SMTP server for its users because the users may not be
online all the time. For dialup users, offloading all their email to an
always (!) available server is preferable than dialing up every 5
minutes to see if the remote mail server is available again. It also
benefits low bandwidth users in that mail for multiple recipients can be
sent to one server rather than each recipient server.
> Why is that none of the mail clients like evolution, are built with
> the kind of logic as sendmail.
Because there is an awful lot of config that can go in an SMTP server,
there is no point burdening each mail client with this facility.
> Why is that users are asked to do one
> more configuration step ( configuring SNTP server for the mail client
> ) that can be eliminated.
Now this is true. I was going to ask "why doesn't DHCP give out SMTP
server info?" but a quick look at the dhcp-options manpage suggests that
it can. My question might be "Which mail clients use this feature? Why
not all?"
--
Paul Black mailto:paul....@oxsemi.com
Oxford Semiconductor Ltd http://www.oxsemi.com
25 Milton Park, Abingdon, Tel: +44 (0) 1235 824 909
Oxfordshire. OX14 4SH Fax: +44 (0) 1235 821 141
to...@aplawrence.com
>Sony Antony <sonya...@hotmail.com> wrote:
>> I found out that sendmail did not use an SMTP server at all. It did an
>Only because you didn't configure a smarthost.
>> MX query, got the mail gateway for the destination address domain, and
>> sent the mail directly.
>> This makes me think why ISPs give an SMTP server and ask you to
>> configure your mail client to point to that.
>Because they generally block outgoing packets aimed at port 25 on other
It hasn't been my experience that is true.
>machines, in order to stop people like you spamming the world. Ask them
>> Why is that users are asked to do one
>> more configuration step ( configuring SNTP server for the mail client
>> ) that can be eliminated.
>(c) because it can't. Stop making wrong assumptions and you will be les
>annoying.
While agreeing with you otherwise, I have to argue this. Email
clients certainly COULD send mail directly rather than using
a fixed server.
And sometimes it is damn annoying that they don't. When I'm
out here in the Berkshires, I use Taconic DSL. Several times
a month, their mail.taconic.net becomes unresponsive. When
that happens, I have to find out what my ip is, and go tell
my aplawrence.com server that it's OK for that IP to
send mail, and then of course tell my mail client to use
aplawrence.com as its smtp. Reboot and I may get a new ip,
so I have to adjust things again. It's annoying, and of
course one solution is to tell the client that THIS machine
is its smtp server and let smtp here handle the outgoing. If
taconic (or back home comcast) blocked outgoing port 25 to
anything but their own servers, I couldn't do that.
There's also other reasons why blocking port 25 would be more
than annoying. I sell and support mail servers, and often need
to verify that somebody is working by making a direct connection
to them with sendmail or just a simple telnet xyz 25. If the isp's
blocked me, that would be impossible.
--
to...@aplawrence.com Unix/Linux/Mac OS X resources: http://aplawrence.com
Get paid for writing about tech: http://aplawrence.com/publish.html
Peter T. Breuer
> Peter T. Breuer <p...@oboe.it.uc3m.es> wrote:
>>Sony Antony <sonya...@hotmail.com> wrote:
>>> This makes me think why ISPs give an SMTP server and ask you to
>>> configure your mail client to point to that.
>>Because they generally block outgoing packets aimed at port 25 on other
> It hasn't been my experience that is true.
Then you should report your ISP immediately to the net cops. Where are
you living? Taiwan? Vietnam? Some other spam center? Oh .. berkshire.
>>machines, in order to stop people like you spamming the world. Ask them
>>> Why is that users are asked to do one
>>> more configuration step ( configuring SNTP server for the mail client
>>> ) that can be eliminated.
>>(c) because it can't. Stop making wrong assumptions and you will be les
>>annoying.
> While agreeing with you otherwise, I have to argue this. Email
> clients certainly COULD send mail directly rather than using
> a fixed server.
They can do. Depends on the client. But you asked why you have to
choose a mail server for the client, and the answer is that clients
really cannot normally access the outside world via port 25 because
outgoing 25 will be blocked by the corporate firewall, except for the
legitimate mail servers for the organisation.
> And sometimes it is damn annoying that they don't. When I'm
> out here in the Berkshires, I use Taconic DSL. Several times
> a month, their mail.taconic.net becomes unresponsive. When
> that happens, I have to find out what my ip is, and go tell
> my aplawrence.com server that it's OK for that IP to
> send mail, and then of course tell my mail client to use
They shouldn't believe you. What makes them think you are not a
fly-by-night spammer? You have a contract to send mail through them?
> aplawrence.com as its smtp. Reboot and I may get a new ip,
> so I have to adjust things again. It's annoying, and of
> course one solution is to tell the client that THIS machine
> is its smtp server and let smtp here handle the outgoing. If
> taconic (or back home comcast) blocked outgoing port 25 to
> anything but their own servers, I couldn't do that.
That is the normal situation.
> There's also other reasons why blocking port 25 would be more
> than annoying. I sell and support mail servers, and often need
> to verify that somebody is working by making a direct connection
> to them with sendmail or just a simple telnet xyz 25. If the isp's
> blocked me, that would be impossible.
And it normally is:
% telnet mx1.inty.co.uk 25
Trying 195.224.93.243...
Trying 195.224.93.241...
no answer, etc. etc. You telling me your ISPs MXs are down? nah ..
must be the universities firewall. Let me log in to my house and
use my telco to get out ...
% telnet mx1.inty.co.uk 25
(nix)
Also blocked on outgoing, via the telco.
Peter
Stefan Zapf
> While agreeing with you otherwise, I have to argue this. Email
> clients certainly COULD send mail directly rather than using
> a fixed server.
>
> And sometimes it is damn annoying that they don't. When I'm
> out here in the Berkshires, I use Taconic DSL. Several times
> a month, their mail.taconic.net becomes unresponsive. When
> that happens, I have to find out what my ip is, and go tell
> my aplawrence.com server that it's OK for that IP to
> send mail, and then of course tell my mail client to use
> aplawrence.com as its smtp. Reboot and I may get a new ip,
> so I have to adjust things again. It's annoying, and of
> course one solution is to tell the client that THIS machine
> is its smtp server and let smtp here handle the outgoing. If
> taconic (or back home comcast) blocked outgoing port 25 to
> anything but their own servers, I couldn't do that.
What about using www.dyndns.org ? Then instead of giving in a specific
ip address, you may use your dyndns address which resolves to the
dynamic ip you are using. This you could do in cases like you described.
cu,
Stefan.
Sony Antony
> Because they generally block outgoing packets aimed at port 25 on other
> machines, in order to stop people like you spamming the world. Ask them
> why they aren't doing it, if they aren't. If they aren't, they will
> likely host a spammer in short order, and have their IP range blocked
> for email by the rest of the world as a result, which YOU won'tlike.
I am with one of the biggest ISPs of US. They dont block. Sometime
back I was with another big ISP, they also didnt block port 25.
I dont understand why this step is neccessary to block spam. relaying
is all that is required as far as I understand.
For instance no ISP should accept emails meant for anything other than
its own domain at any of its server. So xxx.com should only accept
mails for any...@xxx.com. The only exception for this is of course
SMTP servers. But then they have other validation there like making
sure that the IP connecting to it is one of its own.
This is how I thought ISPs usually run. ( AFAIK )
Therefore I dont see any reason for blocking 25 generally.
>
> > Why is that none of the mail clients like evolution, are built with
> > the kind of logic as sendmail.
>
> (a) because tehy are,
Because they are *not*. They always deliver mail to the fixed SMTP
server that you have configured.
(b) because sendmail is not a client.
I didnt say that either. I was saying that if they have been built
with MX record fetching algorithm, they user could have been spared
the pain of configuring the SMTP. ( I am talking about the average
user who buys dummies books to figure out which button to click. For
them its a pain )
>
> > Why is that users are asked to do one
> > more configuration step ( configuring SNTP server for the mail client
> > ) that can be eliminated.
>
> (c) because it can't. Stop making wrong assumptions and you will be les
> annoying.
Did you understand it at all. Because it cant what.
If you are not capable of understanding dont waste your energy on
getting annoyed.
--sony
Rod Smith
[re: ISP policies]
>>Because they generally block outgoing packets aimed at port 25 on other
>>machines
>
> It hasn't been my experience that is true.
Such blocks are far from universal, but they're becoming more common. For
instance, my own ISP (Cox) recently instituted such a block. I gather
that a lot of the big dial-up ISPs do this, but I don't know how common
it is, overall.
--
Rod Smith, rods...@rodsbooks.com
http://www.rodsbooks.com
Author of books on Linux, FreeBSD, and networking
to...@aplawrence.com
>to...@aplawrence.com wrote:
>> Peter T. Breuer <p...@oboe.it.uc3m.es> wrote:
>>>Sony Antony <sonya...@hotmail.com> wrote:
>>>> This makes me think why ISPs give an SMTP server and ask you to
>>>> configure your mail client to point to that.
>>>Because they generally block outgoing packets aimed at port 25 on other
>> It hasn't been my experience that is true.
>Then you should report your ISP immediately to the net cops. Where are
>you living? Taiwan? Vietnam? Some other spam center? Oh .. berkshire.
Nonsense. I live in Massachussetts, and vacation in the Berkshires.
>And it normally is:
> % telnet mx1.inty.co.uk 25
> Trying 195.224.93.243...
> Trying 195.224.93.241...
>no answer, etc. etc. You telling me your ISPs MXs are down? nah ..
>must be the universities firewall. Let me log in to my house and
>use my telco to get out ...
> % telnet mx1.inty.co.uk 25
> (nix)
My DSL isp doesn't block that, thought I think I recall that either
att or comcast might - in which case I simply have to ssh to my own
BSD server to run such tests..
But again, that's no reason for email clients to REQUIRE a smart host.
The fact that your ISP may or may not block is not their concern..
to...@aplawrence.com
>to...@aplawrence.com wrote:
>> While agreeing with you otherwise, I have to argue this. Email
>> clients certainly COULD send mail directly rather than using
>> a fixed server.
>>
>> And sometimes it is damn annoying that they don't. When I'm
>> out here in the Berkshires, I use Taconic DSL. Several times
>> a month, their mail.taconic.net becomes unresponsive. When
>> that happens, I have to find out what my ip is, and go tell
>> my aplawrence.com server that it's OK for that IP to
>> send mail, and then of course tell my mail client to use
>> aplawrence.com as its smtp. Reboot and I may get a new ip,
>What about using www.dyndns.org ? Then instead of giving in a specific
>ip address, you may use your dyndns address which resolves to the
>dynamic ip you are using. This you could do in cases like you described.
I think you are misunderstanding the issues. Dynamic DNS would be useful
to someone who wants http://xyz.org to work where xyz.org is hosted
on a machine using DHCP.
This is a case of relay control. My mail/web server isn't dhcp,
but my home internet access is. Naturally I don't allow
mail relaying from my server, but I CAN tell sendmail that
it's OK to relay if the request comes from a specific ip
address: iow, the dynamic address of my home machine.
Dynamic DNS services don't help that. There are various ways
to automate the procedure, but so far it hasn't been enough
of a need that I've bothered to.
--
to...@aplawrence.com Unix/Linux/Mac OS X resources: http://aplawrence.com
Get paid for writing about tech: http://aplawrence.com/publish.html
>cu,
>Stefan.
Peter T. Breuer
> I am with one of the biggest ISPs of US. They dont block. Sometime
They should .. if any of their clients can send mail to anywhere, then
any of their clients can send spam, and I believe that's against the
law .. they are surely obliged to make sure that nobody is sending out
a million mails a day to random targets! They can only do that if they
at least ACCOUNT who sends what where when.
> Therefore I dont see any reason for blocking 25 generally.
Try again.
>> > Why is that none of the mail clients like evolution, are built with
>> > the kind of logic as sendmail.
>>
>> (a) because tehy are,
> Because they are *not*. They always deliver mail to the fixed SMTP
> server that you have configured.
No they don't. Many clients can and do go straight to the MX if
configged that way. I believe "mail" is one, but I'm not going to
check.
> (b) because sendmail is not a client.
> I didnt say that either. I was saying that if they have been built
> with MX record fetching algorithm, they user could have been spared
> the pain of configuring the SMTP. ( I am talking about the average
The client leaves it to the MTA to choose how to deliver. Generally.
This is a question of separation of tasks.
There are some that can be configured as you would want. I think mail
is one.
>> > Why is that users are asked to do one
>> > more configuration step ( configuring SNTP server for the mail client
>> > ) that can be eliminated.
>>
>> (c) because it can't. Stop making wrong assumptions and you will be les
>> annoying.
> Did you understand it at all. Because it cant what.
The step can't be eliminated. You have to choose your transfer agent.
> If you are not capable of understanding dont waste your energy on
> getting annoyed.
Peter
to...@aplawrence.com
>> I am with one of the biggest ISPs of US. They dont block. Sometime
>They should .. if any of their clients can send mail to anywhere, then
>any of their clients can send spam, and I believe that's against the
>law .. they are surely obliged to make sure that nobody is sending out
Exactly how does a smarthost prevent you from sending spam?
>a million mails a day to random targets! They can only do that if they
>at least ACCOUNT who sends what where when.
People sending that kind of volume don't do it from home machines
and certainly not from GUI mail clients :-)
In fact, the typical spammer trys to find open relays or hijack
home machines etc. to send part of their load out.
>The step can't be eliminated. You have to choose your transfer agent.
It COULD be eliminated. The question is why mail clients don't.
I don't think you've given a good answer to that.
Joe
<p...@oboe.it.uc3m.es> writes
>to...@aplawrence.com wrote:
>> Peter T. Breuer <p...@oboe.it.uc3m.es> wrote:
>
>>>Because they generally block outgoing packets aimed at port 25 on other
>
>> It hasn't been my experience that is true.
>
>Then you should report your ISP immediately to the net cops. Where are
>you living? Taiwan? Vietnam? Some other spam center? Oh .. berkshire.
>
I telnetted into a Spanish mail server this morning from BTOpenWorld,
and yes, I know their dynamic IPs are blocked by some mail systems. Not
this one, and BT certainly weren't blocking outbound.
I had to shift this client onto direct mailing for a few hours a couple
of weeks ago, when their main ISP's SMTP went down. Most servers seem to
accept direct mail from anyone, though I recall an interesting message
during one negotiation:
'This is XXX_Bloggs.com'
'You're bluffing. Reverse DNS inaddr_xxxxx.btopenworld.co.uk'
or words to that effect. It let us in anyway. I don't recall seeing
anyone else running a reverse DNS, though.
My own ISP, Demon, certainly don't block any ports, and are happy for
customers to run SMTP servers and clients. The 'no relaying' is in the
AUP, and I presume the unclued get a stiff email the first time it
happens. I normally run Turnpike on Windows as a client to Demon's
smarthost, but have also run exim/djbdns/fetchmail on occasion.
--
Joe
Paul Black
> My own ISP, Demon, certainly don't block any ports, and are happy for
> customers to run SMTP servers
Happy? Once upon a time you had to run an SMTP server to receive mail,
they didn't have a POP server; at least now you have a choice.
Paul
Sony Antony
> > sent the mail directly.
>
> As it should, sendmail being an SMTP server.
OK
>
>
>
> > This makes me think why ISPs give an SMTP server and ask you to
> > configure your mail client to point to that.
>
> An ISP needs an SMTP server for its users because the users may not be
> online all the time. For dialup users, offloading all their email to an
> always (!) available server is preferable than dialing up every 5
> minutes to see if the remote mail server is available again. It also
> benefits low bandwidth users in that mail for multiple recipients can be
> sent to one server rather than each recipient server.
This is the best explanation so far. Also sounds correct. Thanks a
lot.
>
>
>
> > Why is that none of the mail clients like evolution, are built with
> > the kind of logic as sendmail.
>
> Because there is an awful lot of config that can go in an SMTP server,
> there is no point burdening each mail client with this facility.
You dont have to have all the functionalities of an MTA like sendmail.
As a programmer I think the algorithm should not take more than 100
lines of C code.
1. queries the DNS to get the MX record.
2. sends the email using the SMTP *send* protocol ( No need of
implementing the complete protocol SMTP since it will be used only for
delivery - around 10 pages in RFC 821 ). All the email clients
*already have* this, except that they are always delivered to the
configured SMTP server, which then routs it further.
But I agree from speed point of view, delivering it directly will be
slower, particularly if the destination MTA is far away with high
latency.
>
>
>
> > Why is that users are asked to do one
> > more configuration step ( configuring SNTP server for the mail client
> > ) that can be eliminated.
>
> Now this is true. I was going to ask "why doesn't DHCP give out SMTP
> server info?" but a quick look at the dhcp-options manpage suggests that
> it can. My question might be "Which mail clients use this feature? Why
> not all?"
I didnt know this can be done using DHCP. Thanks. Hmm thinking
further, there needs to be a standardised way ( like /etc/resolv.conf
for DNS with DHCP ) of doing this. Once such files are dynamically
created when the network adapter comes up, mail clients can configure
themselves from it.
Thanks again for the great response.
--sony
Robert Heller
In a message on 17 Jul 2003 07:51:07 -0700, wrote :
SA> >
SA> > Because they generally block outgoing packets aimed at port 25 on other
SA> > machines, in order to stop people like you spamming the world. Ask them
SA> > why they aren't doing it, if they aren't. If they aren't, they will
SA> > likely host a spammer in short order, and have their IP range blocked
SA> > for email by the rest of the world as a result, which YOU won'tlike.
SA>
SA>
SA> I am with one of the biggest ISPs of US. They dont block. Sometime
SA> back I was with another big ISP, they also didnt block port 25.
SA> I dont understand why this step is neccessary to block spam. relaying
SA> is all that is required as far as I understand.
Random factoid: current versions of MS-Windows come with an
*unprotected* SMTP server. It also comes with an un-patched 'personal'
version of IIS as well -- the Code Red Worm and its relatives is very
much still alive.
Lots of spam is now being relayed by unsuspecting MS-Windows users who
have connected their MS-Windows boxes directly to their DSL or Cable
modem. I am still logging Code Red Worm connection attempts to my
Apache/Linux server:
w146.z065104164.chi-il.dsl.cnc.net [65.104.164.146] Thu Jul 17 17:01:49 EDT 2003 /default.ida?XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX%u9090%u6858%ucbd3%u7801%u9090%u6858%ucbd3%u7801%u9090%u6858%ucbd3%u7801%u9090%u9090%u8190%u00c3%u0003%u8b00%u531b%u53ff%u0078%u0000%u00=a
SA> For instance no ISP should accept emails meant for anything other than
SA> its own domain at any of its server. So xxx.com should only accept
SA> mails for any...@xxx.com. The only exception for this is of course
SA> SMTP servers. But then they have other validation there like making
SA> sure that the IP connecting to it is one of its own.
SA> This is how I thought ISPs usually run. ( AFAIK )
SA>
SA> Therefore I dont see any reason for blocking 25 generally.
SA>
SA>
SA>
SA>
SA> >
SA> > > Why is that none of the mail clients like evolution, are built with
SA> > > the kind of logic as sendmail.
SA> >
SA> > (a) because tehy are,
SA>
SA>
SA> Because they are *not*. They always deliver mail to the fixed SMTP
SA> server that you have configured.
SA>
SA>
SA>
SA> (b) because sendmail is not a client.
SA>
SA> I didnt say that either. I was saying that if they have been built
SA> with MX record fetching algorithm, they user could have been spared
SA> the pain of configuring the SMTP. ( I am talking about the average
SA> user who buys dummies books to figure out which button to click. For
SA> them its a pain )
SA>
SA>
SA>
SA>
SA> >
SA> > > Why is that users are asked to do one
SA> > > more configuration step ( configuring SNTP server for the mail client
SA> > > ) that can be eliminated.
SA> >
SA> > (c) because it can't. Stop making wrong assumptions and you will be les
SA> > annoying.
SA>
SA> Did you understand it at all. Because it cant what.
SA> If you are not capable of understanding dont waste your energy on
SA> getting annoyed.
SA>
SA> --sony
SA>
Peter T. Breuer
> Peter T. Breuer <p...@oboe.it.uc3m.es> wrote:
>>Sony Antony <sonya...@hotmail.com> wrote:
>>> I am with one of the biggest ISPs of US. They dont block. Sometime
>>They should .. if any of their clients can send mail to anywhere, then
>>any of their clients can send spam, and I believe that's against the
>>law .. they are surely obliged to make sure that nobody is sending out
> Exactly how does a smarthost prevent you from sending spam?
The smarthost itself keeps records of the spam you send, so that you
can be prosecuted when you send the spam, and have three fingers cut
off for the first offence ...
>>a million mails a day to random targets! They can only do that if they
>>at least ACCOUNT who sends what where when.
> People sending that kind of volume don't do it from home machines
Oh yes they do. It's not that much. You can probably address each mail
to about 50 people without breaking your servers limits, and send one
mail per second, which will get you to a million in about five hours.
> and certainly not from GUI mail clients :-)
?? They use mail spamming software.
> In fact, the typical spammer trys to find open relays or hijack
> home machines etc. to send part of their load out.
Agrred. I.e. they use home machines. They don't have to hijack any, but
they may!
>>The step can't be eliminated. You have to choose your transfer agent.
> It COULD be eliminated. The question is why mail clients don't.
You have to choose the transfer agent and where it is. Some people
don't want to run a sendmail on localhost. Some people can't, because
outgoing 25 is blocked. Others might like it fine.
> I don't think you've given a good answer to that.
Peter
Rod Smith
sonya...@hotmail.com (Sony Antony) writes:
>
> But I agree from speed point of view, delivering it directly will be
> slower, particularly if the destination MTA is far away with high
> latency.
Worse than this is the possibility that the remote site is down entirely,
or completely inaccessible due to a routing issue or whatnot. If the
client were sending directly, it would then either abort too quickly
(assuming the problem is temporary) or have to queue the mail for later
delivery. Queueing the mail might be acceptable if the client were
running at all times, with an always-up Internet connection, but these
aren't safe assumptions.
Rod Smith
> back I was with another big ISP, they also didnt block port 25.
> I dont understand why this step is neccessary to block spam. relaying
> is all that is required as far as I understand.
Spammers could relay through an ISP's mail server, but those servers are
likely to be equipped with fail-safe mechanisms to detect likely spam
runs and shut them down before they can do much damage. This makes it
harder for spammers to send spam than if they could do it directly on
port 25 -- at least, unless the ISP installed some sort of sniffer and
analyzer on outgoing port-25 traffic.
Recently, spammers have begun using Trojans on innocent victims'
computers to relay spam. These programs contact the spammer's system,
collect a load of spam, and deliver it without the knowledge of the
person whose computer has been hijacked. The cases I've heard of like
this all send spam directly via port 25, not via the hijacked computer's
ISP's mail server. Some recent e-mail worms also propagate in this way.
Of course, these programs could be re-written to use an ISP's mail
server, but then it becomes easier for the ISP to detect the problem and
shut it down quickly.
Rod Smith
"Peter T. Breuer" <p...@oboe.it.uc3m.es> writes:
>
> Sony Antony <sonya...@hotmail.com> wrote:
>> I am with one of the biggest ISPs of US. They dont block. Sometime
>
> They should .. if any of their clients can send mail to anywhere, then
> any of their clients can send spam, and I believe that's against the
> law ..
Anti-spam laws are weak or nonexistent in most areas. A few US states do
have such laws, but prosecutions under those law have been nonexistent,
AFAIK. Also AFAIK (and I've watched this carefully), there are no Federal
US anti-spam laws, although several have been proposed over the years.
(Most of the proposed US Federal laws have been very weak, and often in
fact would have legitimized spam had they passed.) I don't know what the
status is of such laws in Spain, though. In any event, a lot of spam
these days is international in nature, which makes tracking it down and
prosecuting it that much harder.
Even if sending spam is illegal in an area, there's the question of
whether the ISP could be held accountable for the actions of their
clients. This is an area of law that's still being worked out as it
relates to the Internet, but my understanding is that ISPs mostly qualify
for "common carrier" protections, at least in the US. Again, I'm
unfamiliar with the standards in Spain (or elsewhere in the world, for
that matter).
Ed Murphy
> Lots of spam is now being relayed by unsuspecting MS-Windows users who
> have connected their MS-Windows boxes directly to their DSL or Cable
> modem. I am still logging Code Red Worm connection attempts to my
> Apache/Linux server:
>
> w146.z065104164.chi-il.dsl.cnc.net [65.104.164.146] Thu Jul 17 17:01:49 EDT 2003 /default.ida?XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX%u9090%u6858%ucbd3%u7801%u9090%u6858%ucbd3%u7801%u9090%u6858%ucbd3%u7801%u9090%u9090%u8190%u00c3%u0003%u8b00%u531b%u53ff%u0078%u0000%u00=a
Is that what that is? I seem to be getting on the order of 1500
hits/month of (a) that, followed by (b) attempts to run something
with ".exe" in it. I just laughed at it, rather than bothering to
figure out *which* IIS exploit it was.
(I do host a handful of legit pages which get a handful of legit
hits, which is why I don't just turn Apache off.)
Peter T. Breuer
> Even if sending spam is illegal in an area, there's the question of
> whether the ISP could be held accountable for the actions of their
> clients. This is an area of law that's still being worked out as it
> relates to the Internet, but my understanding is that ISPs mostly qualify
> for "common carrier" protections, at least in the US. Again, I'm
> unfamiliar with the standards in Spain (or elsewhere in the world, for
> that matter).
Spain has (since this year) internet laws which include legal
recognition of the digital signature and criminality of spamming.
Peter
to...@aplawrence.com
>to...@aplawrence.com wrote:
>> Peter T. Breuer <p...@oboe.it.uc3m.es> wrote:
>>>Sony Antony <sonya...@hotmail.com> wrote:
>>>> I am with one of the biggest ISPs of US. They dont block. Sometime
>>>They should .. if any of their clients can send mail to anywhere, then
>>>any of their clients can send spam, and I believe that's against the
>>>law .. they are surely obliged to make sure that nobody is sending out
>> Exactly how does a smarthost prevent you from sending spam?
>The smarthost itself keeps records of the spam you send, so that you
>can be prosecuted when you send the spam, and have three fingers cut
>off for the first offence ...
Prosecuted?
I don't think any real attempt at anti-spam laws has been yet made, and if
it has, it sure as hell hasn't been tested in the courts, where in
all likelihood it would be shot down instantly.
Peter T. Breuer
> Peter T. Breuer <p...@oboe.it.uc3m.es> wrote:
>>to...@aplawrence.com wrote:
>>> Peter T. Breuer <p...@oboe.it.uc3m.es> wrote:
>>>>Sony Antony <sonya...@hotmail.com> wrote:
>>>>> I am with one of the biggest ISPs of US. They dont block. Sometime
>>>>They should .. if any of their clients can send mail to anywhere, then
>>>>any of their clients can send spam, and I believe that's against the
>>>>law .. they are surely obliged to make sure that nobody is sending out
>>> Exactly how does a smarthost prevent you from sending spam?
>>The smarthost itself keeps records of the spam you send, so that you
>>can be prosecuted when you send the spam, and have three fingers cut
>>off for the first offence ...
> Prosecuted?
> I don't think any real attempt at anti-spam laws has been yet made, and if
Sure it's been made. It's been law for a year here, along with the
recognition of the digital signature, etc. etc. I heard it had become
criminal in the states too, hence all that "[publicity]" stuff in the
spam headers nowadays ...
> it has, it sure as hell hasn't been tested in the courts, where in
> all likelihood it would be shot down instantly.
Laws can't be shot down by courts.
Peter
Robert Heller
In a message on Fri, 18 Jul 2003 03:39:41 GMT, wrote :
"M> On Thu, 17 Jul 2003 21:38:53 +0000, Robert Heller wrote:
"M>
"M> > Lots of spam is now being relayed by unsuspecting MS-Windows users who
"M> > have connected their MS-Windows boxes directly to their DSL or Cable
"M> > modem. I am still logging Code Red Worm connection attempts to my
"M> > Apache/Linux server:
"M> >
"M> > w146.z065104164.chi-il.dsl.cnc.net [65.104.164.146] Thu Jul 17 17:01:49 EDT 2003 /default.ida?XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX%u9090%u6858%ucbd3%u7801%u9090%u6858%ucbd3%u7801%u9090%u6858%ucbd3%u7801%u9090%u9090%u8190%u00c3%u0003%u8b00%u531b%u53ff%u0078%u0000%u00=a
"M>
"M> Is that what that is? I seem to be getting on the order of 1500
"M> hits/month of (a) that, followed by (b) attempts to run something
"M> with ".exe" in it. I just laughed at it, rather than bothering to
"M> figure out *which* IIS exploit it was.
Yep. A GET of "/default.ida?XXXX..." is in fact the 'signature' of
Code Red. It is a classic buffer overrun type of exploit, which is the
reason for all those X's -- the exact number is not random, but it a
measure of the size of the buffer to be overrun. There are some
variations. If a machine is successfully infected, it is then possible
to run the various '.exe' files remotely, which allows for various
second-stage payloads, viruses, or worms.
"M>
"M> (I do host a handful of legit pages which get a handful of legit
"M> hits, which is why I don't just turn Apache off.)
"M>
"M>
to...@aplawrence.com
>> Prosecuted?
They can't?
In this country they certainly can be. Our states and our country have
these "constitution" things, and while our legislators often don't
give a damn about it, now and then our courts do.
Markku Kolkka
> Prosecuted?
>
> I don't think any real attempt at anti-spam laws has been yet made, and if
> it has, it sure as hell hasn't been tested in the courts, where in
> all likelihood it would be shot down instantly.
This may come as a shock to you, but there are actually other countries in
the world than USA, with different laws and different legal systems.
In the EU, article 13 of directive 2002/58/EC says:
"1. The use of automated calling systems without human intervention
(automatic calling machines), facsimile machines (fax) or electronic mail
for the purposes of direct marketing may only be allowed in respect of
subscribers who have given their prior consent.
[...]
4. In any event, the practice of sending electronic mail for purposes of
direct marketing disguising or concealing the identity of the sender on
whose behalf the communication is made, or without a valid address to which
the recipient may send a request that such communications cease, shall be
prohibited."
--
Markku Kolkka
markku...@iki.fi
John Hasler
> I heard it had become criminal in the states too, hence all that
> "[publicity]" stuff in the spam headers nowadays ...
A few states have "anti-spam" laws. There is no Federal law on the
subject. I hope there never is.
> Laws can't be shot down by courts.
Perhaps not in your jurisdiction.
--
John Hasler
jo...@dhh.gt.org
Dancing Horse Hill
Elmwood, Wisconsin
to...@aplawrence.com
>to...@aplawrence.com wrote:
>> Prosecuted?
>>
>> I don't think any real attempt at anti-spam laws has been yet made, and if
>> it has, it sure as hell hasn't been tested in the courts, where in
>> all likelihood it would be shot down instantly.
>This may come as a shock to you, but there are actually other countries in
>the world than USA, with different laws and different legal systems.
True. But we don't care.
No, seriously it is easy to forget that News is worldwide, and that
in itself is often the source of much back and forth, exactly as seen
here.
So, no, it's not a shock, and I honestly wish we had less provincialism,
but I do admit to being guilty of forgetting now and then.
Johan Kullstam
> I was playing with sendmail and .forward file on my RH Linux system.
> When I added an email to the .forward file, sendmail faithfully sent
> it to the destination.
> But then I realized that nowhere had I specified my SMTP server.
From the RFC on e-mail (what is it 2833?)
SMTP client sends mail.
SMTP server receives mail.
Sometimes the server can receive mail for others and relay it to its
final destination.
> So I started investigating using strace ing sendmail.
> I found out that sendmail did not use an SMTP server at all. It did an
> MX query, got the mail gateway for the destination address domain, and
> sent the mail directly.
Sure it did. Sendmail (as client) talked directly to the server at
the other end. This is how e-mail usually works.
You can also relay it through the server at your ISP. Check into
the "smarthost" configuration for your local sendmail.
> This makes me think why ISPs give an SMTP server and ask you to
> configure your mail client to point to that.
Unfortunately, many people have improperly set up systems. This has
led to semi-widespread blocking of people coming from dynamic IP
assignments. Thus you might be pretty well forced to relay through
your ISP's SMTP server.
> Why is that none of the mail clients like evolution, are built with
> the kind of logic as sendmail.
Evolution is a Mail User Agent (MUA) and expects a Mail Transfer Agent
(MTA), e.g., sendmail to do the queuing and retransmission &c.
> Why is that users are asked to do one more configuration step (
> configuring SNTP server for the mail client ) that can be
> eliminated.
Well, if the destination host is down, you might want the mail system
to retry in a few minutes/hours rather than just not working. Usually
the task of ensuring delivery is delegated to a dedicated program for
that task rather than your reader. This lets you choose the pretty
reader independently of choosing the reliable delivery mechanism.
> Any pointers will be appreciated.
> --sony
--
Johan KULLSTAM
Johan Kullstam
> Sony Antony <sonya...@hotmail.com> wrote:
> > I found out that sendmail did not use an SMTP server at all. It did an
>
> Only because you didn't configure a smarthost.
>
> > MX query, got the mail gateway for the destination address domain, and
> > sent the mail directly.
> > This makes me think why ISPs give an SMTP server and ask you to
> > configure your mail client to point to that.
>
> Because they generally block outgoing packets aimed at port 25 on other
> machines, in order to stop people like you spamming the world. Ask them
> why they aren't doing it, if they aren't. If they aren't, they will
> likely host a spammer in short order, and have their IP range blocked
> for email by the rest of the world as a result, which YOU won'tlike.
>
> > Why is that none of the mail clients like evolution, are built with
> > the kind of logic as sendmail.
>
> (a) because tehy are, (b) because sendmail is not a client.
Sendmail is also a client. In this case described above, it is. Read
the RFC.
> > Why is that users are asked to do one
> > more configuration step ( configuring SNTP server for the mail client
> > ) that can be eliminated.
>
> (c) because it can't. Stop making wrong assumptions and you will be les
> annoying.
>
> Peter
--
Johan KULLSTAM
to...@aplawrence.com
>> I was playing with sendmail and .forward file on my RH Linux system.
>> When I added an email to the .forward file, sendmail faithfully sent
>> it to the destination.
>> But then I realized that nowhere had I specified my SMTP server.
>From the RFC on e-mail (what is it 2833?)
>SMTP client sends mail.
>SMTP server receives mail.
And in fact the latest sendmail breaks the client and server into two
different configuration files and usually runs each with different
privileges.
>Well, if the destination host is down, you might want the mail system
>to retry in a few minutes/hours rather than just not working. Usually
>the task of ensuring delivery is delegated to a dedicated program for
>that task rather than your reader. This lets you choose the pretty
>reader independently of choosing the reliable delivery mechanism.
Ding. Good reason (finally).