* Theo v. Werkhoven [09-19-04 07:00]: ...

Not so. If you let an app like kmail do the popping and sending by itself a MTA is not necessary. Only if you want to use a MDA like e.g. procmail then you will also need a MTA.

No, incorrect. You *need* an MTA, SuSE defaults to postfix, or you would not receive system messages, reports, etc. KMail is incapable of handling this. -- Patrick Shanahan Registered Linux User #207535 http://wahoo.no-ip.org @ http://counter.li.org HOG # US1244711 Photo Album: http://wahoo.no-ip.org/photos

* Theo v. Werkhoven [09-19-04 07:39]:

Sun, 19 Sep 2004, by paka@wahoo.no-ip.org:

...

No, incorrect. You *need* an MTA, SuSE defaults to postfix, or you would not receive system messages, reports, etc. KMail is incapable of handling this.

The system messages from e.g. YOU need the sendmail (drop-in) program, but that doesn't have to run as daemon. For the normal sending and receiving of mail a dektop user has no need for a MTA.

agreed -- Patrick Shanahan Registered Linux User #207535 http://wahoo.no-ip.org @ http://counter.li.org HOG # US1244711 Photo Album: http://wahoo.no-ip.org/photos

* Carlos E. R. [09-19-04 09:35]:

The Sunday 2004-09-19 at 14:39 +0200, Theo v. Werkhoven wrote:

...

The system messages from e.g. YOU need the sendmail (drop-in) program, but that doesn't have to run as daemon. For the normal sending and receiving of mail a dektop user has no need for a MTA.

But that still requires an MTA (sendmail, or postfix, which provides a sendmail program) to be installed on the system, even if not loaded permanently.

Yes, he is mixing daemon with MTA. The MTA is still required/installed. -- Patrick Shanahan Registered Linux User #207535 http://wahoo.no-ip.org @ http://counter.li.org HOG # US1244711 Photo Album: http://wahoo.no-ip.org/photos

On Sunday 19 September 2004 4:41 pm, Martin Farmilo wrote:

On Sunday 19 Sep 2004 13:12, Patrick Shanahan wrote:

...

No, incorrect. You *need* an MTA, SuSE defaults to postfix, or you would not receive system messages, reports, etc. KMail is incapable of handling this.

Hmm - this just caught my eye. I've been running Suse for around 6 months now, and had wondered about this issue. Am I supposed to be receiving emails from the system ? If so, is there something special (perhaps in kmail) I need to do to receive them ? Postfix is running on the system, but I've done nothing myself to manually configure it, and indeed can't find anything in the help / admin guide about it. I'd thought about disabling postfix, since it doesn't appear to be serving any useful purpose here, as I fetch / send all my mail direct via my ISPs servers.

Alias root to your email account at your ISP and set your ISP's mail server as the relay_host and you should get any local mail in your normal mailbox. Scott -- POPFile, the OpenSource EMail Classifier http://popfile.sourceforge.net/ Linux 2.6.5-7.108-default x86_64

Mon, 20 Sep 2004, by copperwa11s@yahoo.co.jp:

On Sun, 19 Sep 2004 16:56:19 -0700 Scott Leighton wrote:

...

On Sunday 19 September 2004 4:41 pm, Martin Farmilo wrote:

...

On Sunday 19 Sep 2004 13:12, Patrick Shanahan wrote:

...

No, incorrect. You *need* an MTA, SuSE defaults to postfix, or you would not receive system messages, reports, etc. KMail is incapable of handling this.

Hmm - this just caught my eye. I've been running Suse for around 6 months now, and had wondered about this issue. Am I supposed to be receiving emails from the system ? If

Some messages for 'root' are useful te receive, like those from YOU and some CRONTAB messages (failure to update virus-sig files, logrotate etc.).

...

...

so, is there something special (perhaps in kmail) I need to do to receive them ? Postfix is running on the system, but I've done nothing myself to manually configure it, and indeed can't find anything in the help / admin guide about it. I'd thought about disabling postfix, since it doesn't appear to be serving any useful purpose here, as I fetch / send all my mail direct via my ISPs servers.

Alias root to your email account at your ISP and set your ISP's mail server as the relay_host and you should get any local mail in your normal mailbox.

Hmm.. why not just set it up inside KMail to receive messages from the localhost? That way you wouldn't even to send it outside your pc/network and reveal things that are happening inside your pc.

Kmail can be configured to read from a mailspool afaik, so that should be sufficient to get the local mail. Theo -- Theo v. Werkhoven Registered Linux user# 99872 http://counter.li.org ICBM 52 13 27N , 4 29 45E. + ICQ: 277217131 SUSE 9.1 + Jabber: gurp@nedlinux.nl Kernel 2.6.5 + MSN: twe-msn@ferrets4me.xs4all.nl See headers for PGP/GPG info. +

On Monday 20 Sep 2004 06:33, Scott Leighton wrote:

  Yast -> Network Services -> MTA -> when you get to the incoming mail section, you'll see a button 'Aliases', click it and edit the entry for root. Have the destination point to your email address at your ISP, e.g., usr@sanctum.com. That will cause all mail to root to get forwarded on to your personal email account at your ISP.

  You didn't ask about the relay_host, but I assume you want an answer on that too, on the outgoing mail section (which you'll come to before incoming while you are in the MTA above), put your ISP's smtp server address in brackets, e.g., [smtp.sanctum.com] and if the ISP uses authentication, click the authentication box and fill in the needed information.

Hiya - this is all new to me too. I've been happily using kmail for all my email. Just looked in /var/spool/mail and the root entry in there is huge - presumably tons of mail that's never gone anywhere. I just followed your instructions above - although on the incoming I left blank the bit about incoming SMTP connections. I assumed that was if I wanted to act as an SMTP server on the net, instead of having my internet mail come via my pop mailbox - right ? There was no 'root' entry in the aliases list, but I noticed a box to forward root's mail to an address - this added root into the aliases list. Martin Farmilo

On Sunday 19 September 2004 10:17 pm, Bob S wrote:

...

   Alias root to your email account at your ISP and set your ISP's mail server as the relay_host and you should get any local mail in your normal mailbox.

Scott, please elaborate; Alias root to my ISP ???

Yast -> Network Services -> MTA -> when you get to the incoming mail section, you'll see a button 'Aliases', click it and edit the entry for root. Have the destination point to your email address at your ISP, e.g., usr@sanctum.com. That will cause all mail to root to get forwarded on to your personal email account at your ISP.

Do you mean connect with my ISP and change settings??

No.

How is that possible ??

See above. You didn't ask about the relay_host, but I assume you want an answer on that too, on the outgoing mail section (which you'll come to before incoming while you are in the MTA above), put your ISP's smtp server address in brackets, e.g., [smtp.sanctum.com] and if the ISP uses authentication, click the authentication box and fill in the needed information. Scott -- POPFile, the OpenSource EMail Classifier http://popfile.sourceforge.net/ Linux 2.6.5-7.108-default x86_64

Mon, 20 Sep 2004, by usr@sanctum.com:

On Sunday 19 September 2004 02:49, Jeffrey L. Taylor wrote:

...

Quoting Bob S :

...

Hello SuSE people,

Have a question about Postfix. Back about the 9th or 10th I upgraded KDE to 3.3. That included Kmail which is my only e-mail program.

You need to have some kind of Mail Transfer Agent (MTA) like Postfix on most Linux systems. It looks like it is not correctly setup, though there are other possibilities like your ISP is blocking outgoing mail that bypasses their SMTP servers, a common attempt at reducing SPAM, especially from zombies/trojans. Linux.local may be your machine, check the IP address.

Thanks to all who contributed to this question. Simple question generated quite a large response and as a result I learned quite a bit. Do have a few specific questions from the various responses.

Jeffrey, Sent off an email to my ISP tonight asking about the blocking and/or the POP before SMTP thing. Will let the list know their response.

Theo, you stated:

...

That's not the problem, the problem was a DNS lookup failure according to the bounce message, Postfix couldn't find a place to drop the mail because the DNS didn't give it one, and so it returned the mail to sender. Please elaborate. Where and why couldn't Postfix find the DNS. The DNS for my ISP ? or for the recipient? And if so how do I check it? The DNS for my ISP is defined in PPP someplace.

It wasn't that Postfix couldn't find a DNS, but the DNS didn't answer the query that the DNS received from Postfix, and so POstfix didn't know where to drop off the mail. That's the way mail servers work: a user wants to send a mail to user@example.com. Postfix gets the mail from the Mail User Agent (the mail client program like e.g. kmail), looks at the recipient address (example.com) and tries to find out which mail host is responsible for the example.com domain. Postfix does this by first asking root DNS servers for what DNS knows anything about the .com top level domain, then it asks the first responding server what DNS knows about example.com domain, and if Postfix got that address it asks example.com what host receives the mail for example.com. If Postfix finally knows the answer (e.g. mail.example.com) then it can send the mail on to its destination. If one of the DNS server doesn't answer (usually the one responsible for the domain), the request from Postfix times out and the mail never gets send. In your case the DNS responsible for .com tld didn't know esupport.com, so the query ended there. You can check with 'dig -t mx example.com' if your 'resolver' can find the Mail eXchange record for example.com.

Now, if I read the posts from Theo, I do not NEED an MTA - OK - understood ??

Not one running as daemon, if all you do is Desktop stuff, no.

But I DO have Postfix. According to Patrick Postfix is needed internally by SuSE for things like mail for root etc. Good- OK Carlos says that Kmail actually uses Postfix to send it's mail. ( required?) And Patrick agrees, stating: >"Yes, he is mixing daemon with MTA.  The MTA is still required/installed." ----- Or, only if it is there to use? But Kmail can actually do it all by itself? A little confused here !!

Kmail has it's own SMTP engine, and doesn't need an external mail server like e.g. Mutt. You can set your ISP's mail gateway address in kmail and drop the mail directly into their care. For system messages you wish to receive you do need a MTA though. Theo -- Theo v. Werkhoven Registered Linux user# 99872 http://counter.li.org ICBM 52 13 27N , 4 29 45E. + ICQ: 277217131 SUSE 9.1 + Jabber: gurp@nedlinux.nl Kernel 2.6.5 + MSN: twe-msn@ferrets4me.xs4all.nl See headers for PGP/GPG info. +

On Monday 20 September 2004 14:50, Carlos E. R. wrote: ..............<snipped all kinds of stuff in this message>.................

The Monday 2004-09-20 at 01:17 -0400, Bob S wrote:

...

The DNS for my ISP is defined in PPP someplace.

There are two addresses but assignment is done dynamically.

Local mail, sent by Yast when installing a package, or by a cron job when it has problems, or when hylafax receives a fax, etc, all these need a local MTA.

...

...

   Alias root to your email account at your ISP and set your ISP's mail server as the relay_host and you should get any local mail in your normal mailbox.

No, no, don't do that. Alias root to a local user. Scott Leighton explained how to do that in Yast, but use your local user name instead, not the address your ISP gave you.

OK, will do that (if I can figure out how to do it ) ( Think I need to give it a real name instead of linux.localhost)

...

Now, on to my final understanding:

Carlos wrote: As I said, Kmail needs an SMTP server. It can be your local postfix or sendmail or qmail, or it can be another one on your intranet, or extranet, at your ISP, or at the other end of the world. It doesn't matter. It needs one, but you are not required to provide one on your machine.

OK, ....so smtp.sanctum.com would suffice??

...

And were the messages received by the server?

...

Your local machine server, yes. Remote or ISP server, no.

OK - are you saying my local server on my machine? That nothing was accepted by the ISP and that there was/is no record of that mail message at the ISP??

Perhaps I had a little mistake in reading the message. The sequence is as follows:

Read carefully several times if needed: it is not so easy to follow if you are not habituated to this things. Not even for me ;-)

Using KMail/1.7 you send an email to "eSupport.com" <Flashupgrades at esupport.com>, from Bob Stia <rnr at sanctum.com>, and dated Sun, 12 Sep 2004 21:43:35 -0400.

It is handled by postfix at linux.local, 3 seconds later.

The Fri, 17 Sep 2004 22:17:25 -0400 (EDT), postfix gives up (exactly five days later, plus half an hour), and sends an email notifying this fact, from MAILER-DAEMON@linux.local to rnr at sanctum.com. The reason given is a DNS failure. That is, for whatever reasons, it could not find the IP number of the destination machine at esupport.com. This is the only problem. It can be a misconfiguration at your side, or a network problem of some sort. It can even happen if the machine is unpowered, or without network connection, during those days. Or because you double boot to windows, I can't know.

Notice that I'm following the "Received" headers, from bottom up - the most recent one is at the top of the message. I now analize the last (top) one:

Received: from linux.local (pool-56.max4.marlowe.net [64.58.*.*] ) by SANCTUM.COM with ESMTP (IOA-IPAD 4.03g/96) id 75MX400 for <rnr at sanctum.com>; Sat, 18 Sep 2004 02:10:39 -0400

As I said, the report is sent to the original sender of the message, that is rnr at sanctum.com, ie, you. That address is external. The "received" header is read backwards (this is customary):

The SMTP server at "SANCTUM.COM" (with software named "IOA-IPAD 4.03g/96") handled a message with ID "75MX400", that whas received from a machine that calls itself "linux.local", but that in fact is "pool-56.max4.marlowe.net", with IP such and such. This happened on "Sat, 18Sep 2004 02:10:39 -0400", that is, around four hours later that when postfix created the "return to sender" message we are seeing. This four hours delay might be because you were not connected during that interval.

I assume that the message was now handled to the POP server of your ISP. This is not seen on the headers, I suppose because you clipped them.

No, I did not clip anything.

Finally, you fetch it back, read it, and start typing questions at us :-p

It is possible to configure postfix so that it knows that rnr@sanctum.com is you, and send it to your local user directly. Another day.

Is it?? Yes, I hope so. Have to figure out how to do that. That would avoid all of that sending to the ISP and back thing. Right??

...

See, that is what confuses me. I download messages from my ISP and receive these "undelivered" messages. Are they not coming from the ISP?

For this message we are analyzing, yes, the final hoop is through your ISP.

...

Coming from my "local" server?

Yes as well - the initial part.

...

Are you saying my machine is holding the messages for six days and then notifying me they are undeliverable??

Yes. When it finally gives up, the message is returned to the sender, and thinking it is an external address, it is sent out, and you finally see it coming from your ISP account. But the "return to sender" was not generated there, nor was it delayed there. All that happened on your machine.

Yes, because the reason is reported thus: | Diagnostic-Code: X-Postfix; Host or domain name not found. Name service | error for name=esupport.com type=MX: Host not found, try again

Notice the "try again" part --------------------------^^^^^^^^

It is thought to be a temporary problem, and postfix obeyed, it tried again, and again, and again. You know, computers are very patient. They can not get angry or tired: they try and try and try till told to stop :-p

Or told another way, they try for the preprogrammed time lapse, five days or whatever. If the problem had been reported to be permanent (permanent failure, non existent domain, for example), it would have given up immediately.

Notice that some recipient machines take an antispam measure that consist in giving a temporary failure. I know they exist, I can not say that this is the case here.

Don't know for sure. This ISP is very strong on the "spam" thing though.

You could grep the mail log to find all instances of "3D54223317", that is the ID of the original message that could not be sent, to find out how many times it tried, and the reasons given each time. Look at '/var/log/mail' or '/var/log/mail.debug', depends on your setup.

...

...

...

even though it was stated that my messages were refused? and just now returned? and why if they had valid delivery addresses?

No, they were not refused, nor rejected. They were indeed returned because of unknown address.

Yes, but when??? When I sent them, or were they actually on the ISP server trying for six days?

No, on your local machine (linux.local). Your local machine was trying for 5 days and failing. See above for the times and dates.

You can easily find out if this is happening right now. Type:

mailq

Showed nothing. Even though last night I tried sending a test message. ( when I was verifying the POP before SMTP authentication)

...

...

Look carefully:

...

Subject: Undelivered Mail Returned to Sender

        See? Returned.

Yes, again, but to who? my local server or to the ISP for resending ?

They are returned to you, the human person responsible, because you are the originator of the email, by your local postfix server. Just as a paper mail with an unknown destinatary - but here time outs are measured in months :-)

It so happens that the return address is not recognised as local (your machine is named linux.local), so it is sent outsid. There, your ISP collects it and put it on your mail account.

OK, then I need to make Postfix recognize my local machine. Have to figure out how to do that.

...

...

...

X-Spam-Checker-Version: SpamAssassin 2.63 (2004-01-11) on linux.local X-KMail-EncryptionState:  

        (spam check done on receipt, by kmail)

This is the Postfix program at host linux.local.

     It is your local machine who is speaking, not the ISP.

...

...

I'm sorry to have to inform you that your message could not be be delivered to one or more recipients. It's attached below.

        Main problem: it could not deliver for some reason.

...

For further assistance, please send mail to <postmaster>

        Notice that _you_ are the postmaster, so you have to give further       assistance to the user - also you :-p

If, I assume what you are saying, is all on the local machine, where would I find these files??

What files?

Where Postfix keeps all of this stuff.

...

...

        And the problem was:

...

<Flashupgrades at esupport.com>: Host or domain name not found. Name service  error     for name=esupport.com type=MX: Host not found, try again

        See? It could not find the host (machine) named "esupport.com".           Notice the "try again" part: so it did, for six days, before         giving up and returning the mail to the sender.

Well, it never would. right?? Because it is sitting on my machine without access to the internet.

...

Again, from where? is it trying, internally on my own box, or out there on the ISP??

Internally. At the machine identified as "linux.local", that is, your machine.

As above

...

...

...

Would like to know about this and should I be disabling Postfix somehow?

Disable Postfix? Why? postfix is working correctly.

Now, why did it say that "esupport.com" does not exist? That's a diferent problem. Try this on your machine that has the local postfix server:

Probably because of the blocking or pop before SMTP thing. As you said. "a different problem" (but certainly related though)

No, POP before SMTP has nothing to do here, for this particular email. It did not get out.

Couldn't it be that it didn't get out by being refused by the ISP because of the "POP before SMTP" thing??

...

Anyway, thanks to all who contributed. Let's see what my ISP has to say about this.

In this case, probably not much, not their fault. Depends on how helpful they are, and how complacent with people running their own smtp servers (postfix) on linux.

The only problem you have to solve is to find out why the internet name was not found.

Because the ISP wouldn't accept it?

Why the DNS queries failed. Don't know ! /related t not accepting it? I need to know more about your setup and what you did in those days with your machine plus networks - you know, Crystal balls are hard to get by, and they are out for repairs continuously :-P

I think, and jump on me if I am dead wrong, the internet name was not found because the ISP would not accept it. ie: "the POP before SMTP authentication" thing. Following are excerpts which I sent to the ISP and their reply. -----------------------------------------------------

...

Are you using POP before SMTP authentication ?? I assume you are also blocking mail sent outside your SMTP server.

They replied:

Either POP before SMTP authentication or straight SMTP authentication will work fine.

So, ??? so what???? To check this out I tried to send by sending first - rejected- then did download of mail - sent again - went through. Seems that I cannot send mail until after I fetch mail.

(ie, more in your logs and your head)

-- Cheers, Carlos Robinson

Thanks Carlos, you are a gem !! Appreciate the patience, understanding, and wilingness to solve this. (educate me to understand) Bob S. PS My crystal ball will be returned shortly after all upgrades and maintenace have been performed. Will send it to you ASAP so you might continue in solving problems like this for us. :-)

On Tuesday 21 September 2004 19:39, Carlos E. R. wrote:

The Tuesday 2004-09-21 at 02:05 -0400, Bob S wrote: ............... Hope that this has not made it too confusing though.

First of all at end of the last message Carlos asked about my system. (No crystal ball) Anyway 8.2 with stock kernel 2.4 KDE 3.3 upgraded not too long ago which has Kmail 1.7. That is when problem with the outgoing mail started. I have never ever set up a MTA (since SuSE 6 something) Always used either Mozilla or Kmail. Kmail being my favorite and used now for several years. When I received the returned mail and saw Postfix and linux.localhost it surprised me and prompted my question for this thread. I do remember though back about 7.2 or somewhere therein, I used to get root messages in my mailbox. Have not now for quite a long time. OK - Back to basics - Configuring Kmail -Let's forget about Postfix for now!. Kmail - configure - Network - sending - When I set it for smtp.sanctum.com When attempting to send a message I receive a message that the server is unknown - When set for mail.sanctum.com or just sanctum.com receive a message that the address for the recipient is unknown. When authentication for either of those addresses is added - Get message that authentication failed, most likely the password. (password IS correct.) Now, that being the case, and thinking that Postfix is somehow involved, (due to info in the returned mail) went into YAST and configured my MTA (Postfix?) Kind of repeated what was configured in Kmail. (including authentication)(as a little further down) An aside: The second message from me to my ISP and their reply: -----------------------------------------------------

...

Tried it out by sending first - rejected- then did download - sent again - went through. Seems that I cannot send mail until after I fetch mail.

SMTP Auth is apparently failing, but POP before SMTP is working.

OK - Now back to configuring Postfix Carlos said:

...

No, no, don't do that. Alias root to a local user. Scott Leighton explained how to do that in Yast, but use your local user name instead,

...

not the address your ISP gave you.

Went into YAST - networks - MTA - Couldn't find anything as an "alias root" So, under the incoming mail section; server, I put sanctum.com, under Remote user name, i put bob@linux.local Under local user, I put bob Think I may have set it up correctly ?

Yast does it, and I suppose it is explained on the printed manual. Any way, it is point and click. As I said, Scott explained it, just use a local name.

As per above

(me)

...

...

...

Now, on to my final understanding:

(Carlos)

...

...

As I said, Kmail needs an SMTP server. It can be your local postfix or sendmail or qmail, or it can be another one on your intranet, or extranet, at your ISP, or at the other end of the world. It doesn't matter. It needs one, but you are not required to provide one on your machine.

(me)

...

OK, ....so smtp.sanctum.com would suffice??

(Carlos)

Yes, if you like it, you can use it.

No, didn't mean that. I was referring to what should be the address for the external server if I didn't have a local server.

Or you can use localhost, which means your postfix will handle it.

OK, as per above in the YAST setup. (Hope it was done correctly) Now, point of order: Both Kmail and Postfix have been set up. Does one take precedence over the other ??

The advantage (and disadvantage) of using a local smtp server for sending to internet is that you have more control over it. Your machine would be sending directly to every body you write to, without using an intermediary server under someone else's control.

The problem is that some recipients reject them.

Also, I think, not sure, that my ISP blocks them and I must use them as a relay.

...

...

I assume that the message was now handled to the POP server of your ISP. This is not seen on the headers, I suppose because you clipped them.

No, I did not clip anything.

And you clicked on full headers? If that is so, something weird happened.

What I posted was "View source" ( Seems that in 1.7 that this has replaced the "headers" when you right click on the message) ( seems to show everything) and I copied it word for word.

...

...

Finally, you fetch it back, read it, and start typing questions at us :-p

Yeah !!! How true, Have to call on the experts when I have no idea as to what is going on. :-(

...

...

It is possible to configure postfix so that it knows that rnr at sanctum.com is you, and send it to your local user directly. Another day.

Is it?? Yes, I hope so. Have to figure out how to do that. That would avoid all of that sending to the ISP and back thing. Right??

Right. It is very simple. As root: nimrodel:~ # cd /etc/postfix/ nimrodel:/etc/postfix # mcedit virtual (or your favorite editor)

add the line:

rnr at sanctum.com bob

Did not try that as I am not sure what that will do, Is that a manual method in lieu of setting up Postfix in YAST?? Will do anything at this point. Just like to know what is happening and why.

and save. Replace the at by @, of course. I assume "bob" is your local name. Finally:

Correct about bob

nimrodel:/etc/postfix # postmap virtual nimrodel:/etc/postfix # rcpostfix reload Reload mail service (Postfix) done nimrodel:/etc/postfix #

and to double check:

nimrodel:/etc/postfix # less /var/log/mail

go to the end of the log file, you should see something like:

Sep 22 00:36:15 nimrodel postfix/master[5233]: reload configuration

and a few more lines if there are pending mails (or problems), like:

Sep 22 00:36:15 nimrodel postfix/qmgr[12185]: 28D0720D17: from=, size=2124, nrcpt=1 (queue active) Sep 22 00:36:35 nimrodel postfix/smtp[12188]: 28D0720D17: to=<suse-linux-s at suse.com>, relay=none, delay=8438, status=deferred (Host or domain name not found. Name service error for name=suse.com type=MX: Host not found, try again)

Then send and email to yourself, and check, with command "mailq". By the way, if you want to delete and email on the mail queue, once you know its ID, the command is "postsuper -d MAIL_ID_HEX

Notice that the queued email above is giving right now a DNS error, try again. That's normal, I'm not connected.

...

...

You can easily find out if this is happening right now. Type:

mailq

Showed nothing. Even though last night I tried sending a test message. ( when I was verifying the POP before SMTP authentication)

Then it was sent.

No, it wasn't - The message was that the mail would be held in the outbox until the problem was resolved or I deleted it. mailq showed nothing even though right now I have two mails waiting to be sent which had been rejected for as per the three examples I posted near the top of this message. As stated in the above paragraph, tonight I composed an email and tried to send it at least a half dozen times while trying different configurations in Kmail and Postfix. mailq showed nothing.

Notice that postfix is not affected by POP before SMTP of your ISP, unless you use a relay server, because it is not sending to you ISP (as a relay). It sends directly to the destination, bypassing your ISP and not asking for permission.

Unless the ISP is blocking? right? And, mail.cf says that sanctum.com is my relay server.

However, if your postfix is configured to relay all your mail to your ISP SMTP server, your ISP will request some type of authentication.

But as per way above, the message is stating - authentication failed -Perhaps an incorrect password ??........ ( NO, password IS correct)(It is the same as what logs me on to the ISP )

Postfix can not handle POP before SMTP. Instead, the passwords are stored in /etc/postfix/sasl_passwd, where domain would be your ISP domain:

domain username:password

OK - that is now changed - was as above and now contains the required username and password

(with a "postmap sasl_passwd" command to apply it) and a configuration change is done in the main.cf file:

smtp_sasl_auth_enable = yes smtp_sasl_security_options = smtp_sasl_password_maps = hash:/etc/postfix/sasl_passwd smtpd_sasl_auth_enable = no

(last line set to no, because your postfix does not receive).

I have only two of the above: smtp_sasl_auth_enable = yes (which was set to no and changed to yes) and: smtpd_sasl_auth_enable = no (which is the same and left it alone) But, I did a SuSEconfig on postfix and I got this message: rebuilding /etc/postfix/sasl_passwd.db *** WARNING *** Found /etc/postfix/main.cf.SuSEconfig, exiting... *** WARNING ***

I don't think you need all this, I just explain for completion, and because of the time lapse between you and me. Just in case.

I appreciate that. And this is not a horrible must solve problem because I can just send a request to download POP, and then send the queued messages in the outbox. It is just very aggravating and should work properly.

A bit of background.

Postfix (as any SMTP server) is designed to send directly to destinations. If you send an email to somebody@somemachine.somedomain, postfix first finds the IP number of that machine, using the DNS, and then talks directly to that machine, and transfer the email. The "somemachine.somedomain" machine should accept all email that is directed to its own users, it can not ask for authentication (this "might" not be so in some cases, but I don't know much about that strange case). It should reject mail for anybody else, unless you are a user of that machine.

OK - good. When I spoke with my ISP I explained that I could send messages within it's domain, but any messages outside it's domain failed. The reply was, (after sending example messages ) "It appears as though you are trying to send messages on your own" Not understanding what they had said at that moment I said I would check it out.

Instead of that, it

(Meaning Postfix ??)

can hand off all email to an intermediary, for example smtp.yourisp.domain. Your ISP is acting as a "relay server". It must authenticate you, or else it will become a spammer sanctuary or nest.

It appears that is true in my case. Do some ISP's require their accounts to use them as a relay server?

There are many methods for that, and user/password pairs is very flexible, IMO.

Much more about all this stuff can be found at the howtos at "/usr/share/doc/howto/en/txt/":

Mail-Administrator-HOWTO.gz Mail-Queue.gz Mail-User-HOWTO.gz

Will read it all.

I hope I have not confused you further :-)

Yes and no, you have certainly educated me on how this works. {In last message I ask about Postfix files }

Ok, they are somewhere under "/var/spool/postfix/*":

. .. active bounce corrupt defer deferred flush hold incoming maildrop pid private public trace

Don't touch that. To know what mails are in there, and why, use "mailq". If you need to hold, delete mails, use postsuper. If you want to have a look, use "mc" to browse files. If you want to read a mail that is there, use "postcat file".

OK, for curiosity's sake, I went and browsed those directories.Out of the hundred or more directories, there is not one mail in there. 0 -nothing - only contains one pid file of 17 bytes.

For example, "postcat -q 28D0720D17|less" displays my queued email (one) by that ID.

Couldn't do that, nothing there. (have two messages in the outbox of Kmail though)

Postfix has a good documentation in /usr/share/doc/packages/postfix/html/index.html, including a FAQ.

OK, will read that also.

...

...

No, POP before SMTP has nothing to do here, for this particular email. It did not get out.

Yes, but it could be that if failed smtp authentication and did not get out, right ??

...

No. Unless you have configured postfix to relay to your ISP, and that is not the default config. It could be, but I doubt it.

But, when I looked at main.cf it showed relay host=sanctum.com

...

...

Why the DNS queries failed.

Don't know ! /related to not accepting it?

I think, and jump on me if I am dead wrong, the internet name was not found because the ISP would not accept it. ie: "the POP before SMTP authentication" thing. Following are excerpts which I sent to the ISP and their reply.

No, no. A DNS query can not be denied. The problem would be reported differently (no route to host, for example, or rejected for a reason).

Didn't you say your machine was not connected?

No, it was connected each time I attempted to send a mail.

...

----------------------------------------------------- Message I sent to my ISP:

...

...

Are you using POP before SMTP authentication ?? I assume you are also blocking mail sent outside your SMTP server.

They replied:

...

Either POP before SMTP authentication or straight SMTP authentication will work fine.

---

So, ??? so what???? To check this out I tried to send by sending first - rejected- then did download of mail - sent again - went through. Seems that I cannot send mail until after I fetch mail.

Then configure the sasl file as I said above. Only needed if relaying. Could be the case.

Made those changes as suggested - As per above - Attempted to send a message and got an "authentication failed" message. Don't know what to do next. Guess I will be downloading mail before I can send out new mail. Here is my ISP's last message to me: The mail server should be mail.sanctum.com or just plain sanctum.com. "If SMTP authentication isn't working it would be because your system isn't sending the smtp login/password correctly.  That's sort of odd since you are clearly sending the POP before SMTP correctly." Soooooo..... I guess there is something wrong in my confirguration someplace. Darned if I know what or where, or could it be some kind of bug in 1.7 ?? That is why I asked apout what takes precedence, Kmail or Postfix, Since I think that Postfix is now properly configured. Just don't know.

Thanks to all, and especially to you Carlos, who contributed to this thread. Must have been at least 50 replies. Bob S.

On Thursday 23 September 2004 07:59, Bob S wrote:

On Tuesday 21 September 2004 19:39, Carlos E. R. wrote:

...

The Tuesday 2004-09-21 at 02:05 -0400, Bob S wrote:

............... Hope that this has not made it too confusing though.

Time ago the sign '>' was preceded with the initials of each correspondent. That method is customary in fidonet, for example. Pity current email clients do not support it well. By the way, I'm now using kmail in order to see how it is configured. Normally I use Pine - this demonstrates that it is possible to use a number of mail clients in Pine, sharing the folders.

First of all at end of the last message Carlos asked about my system. (No crystal ball) Anyway 8.2 with stock kernel 2.4 KDE 3.3 upgraded not too

I was more interested in your network configuration. One computer or more, etc. But I think I don't need it now. Also I referred to the fact that looking at the /var/log/mail file a lot can be known about what is the problem/s. That is important.

long ago which has Kmail 1.7.

I have suse 9.1 with defaults kde 3.2.1 and kmail 1.6.2

That is when problem with the outgoing mail started. I have never ever set up a MTA (since SuSE 6 something) Always used either Mozilla or Kmail. Kmail being my favorite and used now for several years. When I received the returned mail and saw Postfix and linux.localhost it surprised me and prompted my question for this thread. I do remember though back about 7.2 or somewhere therein, I used to get root messages in my mailbox. Have not now for quite a long time.

Yes, you need the alias thing. Also, the default 'Inbox' of kmail does not read from system mail, it has to be added. Ok, instead of answering every question as they come, I'll try to clarify some of them. Read the following carefully. **** To avoid the error you were getting with suseconfig. ======================================= My fault. I told you to edit postfix configuration files directly, and that has that side effect. You can undo the changes: nimrodel:/home/cer # cd /etc/postfix nimrodel:/etc/postfix # ls *SuSEconfig main.cf.SuSEconfig master.cf.SuSEconfig nimrodel:/etc/postfix # You see the files 'main.cf.SuSEconfig' and 'master.cf.SuSEconfig'. Those are the files SuSEconfig wants to activate, but has stopped short of doing it because it noticed _I_ modified 'main.cf' and 'master.cf' manually. Therefore, you choose: let suseconfig do its work, or do it yourself. I think you'd better choose suseconfig way, at least till you are confortable. Just copy the '*SuSEconfig' files over the modified '*' file. You'd better backup it before just in case. Then, we go to Yast, and configure mail the easy way: **** Yast ============ Start "yast", go to "Network Services/ Mail Transfer Agent". Connection type : permanent. Enable Virus Scanning: you choose. Next Outgoing mail server None Masquerading - empty Authentification - empty Incoming mail - I'll paste it (use fixed width font to view): ┌Incoming Mail───────────────────────────────┐ │[ ] Accept remote SMTP connections │ │ ┌Downloading─────────────────────────────┐ │ │ │Server Protocol │ │ │ │▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ AUTO▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ │ │ │ │Remote user name Password │ │ │ │▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ │ │ │ │Local user │ │ │ │▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ │ │ │ │ [Details...] │ │ │ └────────────────────────────────────────┘ │ │Forward root's mail to │ │bob▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒│ <==== │Delivery Mode │ │Through procmail▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ │ │ [Aliases...][Virtual domains...] │ └────────────────────────────────────────────┘ Don't define the downloading section, because you use KMail for that. Leave it empty. In virtual domains, you can define: your_email@yourisp.hisdomain bob In aliases, there will be many, but one of them will be: root bob which means that mail sent to 'root' will instead goes to 'bob'. Finally,type Finish. **** Some testing ================= For these tests, we will use the system mail program, ie, a CLI. The program 'mail' in Suse 8.2 and later is in fact 'nail'. Check it, typing 'man mail' will bring up nail's man page. It is a symlink. Do "sux -" on an xterm, or open a root console in kde's Konsole. Type: nimrodel:/etc/postfix # mail cer Subject: test hola . <==== the dot\n signals the end of the email. EOT nimrodel:/etc/postfix # nimrodel:/etc/postfix # You have new mail in /var/spool/mail/cer nimrodel:/etc/postfix # That is, send a manual email to yourself. If after a minute, when typing enter on bob's console (on any bob's console, in fact), you see the message that you have a new email, all is well - so far :-). You may try to send another to 'root', it should also appear there. You may or you may not see them using kmail, that is a different matter. Later. You may - no, do - check it looking at the '/var/log/mail' file. You will see entries from several postfix's processes, and perhaps amavis and/or spamassassin. The end will be something like: Sep 24 00:00:05 nimrodel postfix/local[9618]: 35DAA20D14: to=, orig_to=, relay=local, delay=1, status=sent (delivered to command: /usr/bin/procmail) Sep 24 00:00:05 nimrodel postfix/qmgr[5286]: 35DAA20D14: removed Now, we try sending an email to "your_email@yourisp.hisdomain". nimrodel:/etc/postfix # mail your_email@yourisp.hisdomain Subject: Hello there! See me? . EOT nimrodel:/etc/postfix # You have new mail in /var/spool/mail/bob It will show in the log: Sep 24 00:09:32 nimrodel postfix/pipe[9797]: C765820D11: to=, orig_to=, relay=vscan, delay=1, status=sent (nimrodel.valinor) ... Sep 24 00:09:32 nimrodel postfix/local[9802]: 341CD20D14: to=, relay=local, delay=0, status=sent (delivered to command: /usr/bin/procmail) Notice the first line of the two shows how it is redirected to a different address, ie, instead of going to your "external" address, it goes to your local one. **** Advanced settings, postfix. ================================= If you need to tailor postfix settings without disturbing yast or suseconfig, just edit the file '/etc/sysconfig/postfix' instead, and later call suseconfig. Else, go to Yast/system/"/etc/sysconfig editor" --> Network/Mail/General Network/Mail/Postfix The entries are commented. In the '/etc/sysconfig/postfix' file, it is even possible to enter new settings; for example: POSTFIX_ADD_MESSAGE_SIZE_LIMIT="10240000" POSTFIX_ADD_BOUNCE_SIZE_LIMIT="4800" Trick: If you need verbose debug info in postfix, edit this in 'etc/postfix/main.cf': debug_peer_level = 2 debug_peer_list = sanctum.com With this, conversations between your postfix server and that server will be logged VERY verbosely. Notice that if you have problems only with one remote server, you only need debug info for that one. **** Sending outside. ======================== The settings I explained above are for postfix sending on its own (what your ISP said you might be trying to do). It works, unless some server thinks you are a spammer. So far, we have only tried to send locally. We can now try to send to an external address; if you have a secondary address (or a friend's - you can use me, but the delay is long), try it: mail -r "bob@fromaddress" bob@toaddress" Now, this may, or may not work. If it doesn't, the reason will be seen on the log file. However, it doesn't matter if your provider requires pop before smtp or whatever authentication, because you are bypassing them. Your DNS setting, however, does matter. Ok, so far, so good. If it doesn't work, I have to see the log. **** Configuration possibilities. =================================== 1) Postfix may send directly, on its own, like above. The advantage is that you control it. The main disadvantage is that some servers outside will not accept your email. Depends on your case; for me it works, and in fact, I have to use it (easier method when having different accounts). For some people it may be very problematic. 2) Postfix may send to another SMTP server, who will then send the email. This is a relay server. In this mode, we are not using the full capabilities of postfix, but it also has some advantages. In some cases, it can be a necessity, because the corporate network or the ISP has a firewall that blocks outgoing SMTP connections. In my opinion, it makes sense on a business (more control, no secrets given), but it is arguable for a provider (it makes life difficult for spammers, but you might want to use another server for your own reasons, like having several accounts). 3) Your MUA (Kmail) may hand off mail to the localhost smtp server. If postfix is already configured, it makes life easier, all MUAs just use the local services. If you have a real domain name and several local users, this is the way to go, certainly. For a single account, it may be an overkill. It also depends on your preferences; this is the method I always use, and some people never use it. ¡You choose! :-) 4) Your MUA (Kmail) may hand off mail to _any_ remote smtp server (unless your ISP blocks it). Limitations: ------------- In case #4, options #1 and #2 do not matter. In case #3, if there are problems, the log file will tell me what exactly is going wrong. I can even increase verbosity as needed. With kmail (#4), my toolbox is more limited. In cases #2 and #4, you need to setup authentication. How? In case #4, some MUAs can first fetch mail, then send. However, postfix (#2) can not handle it. Otherwise, #2 and #4 should use login/password authentication. Auth #2 - For postfix, I already explained it. The login/pass pairs go into file '/etc/postfix/sasl_passwd': mail.remote.server user:pass After editing postfix databases files (like this), you need to update the binary database: postmap sasl_passwd and reload postfix (I think it will reload itself after sometime): rcpostfix reload Auth #4. Go to Settings/Configure Kmail, then Network, Sending tab. Chose modify or new (notice: you may set up several outgoing accounts: one may be local, another remote). You will have a dialog showing "Transport: smtp". The "general" tab will have something like: name: arbitrary host: mail.sanctum.com (or sanctum.com, as they say) port: 25 X server requires authentication login whatever password whatever X store SMTP password in configuration file (or it will ask for it every time) On the security tab, press the button "check what the server supports", it will tell you what to use. *** Well... what else? ======================= I don't know, I hope this write up is prety much complete O:-)

OK - Back to basics - Configuring Kmail -Let's forget about Postfix for now!. Kmail - configure - Network - sending - When I set it for smtp.sanctum.com When attempting to send a message I receive a message that the server is unknown - When set for mail.sanctum.com or just sanctum.com receive a message that the address for the recipient is unknown.

Notice that this last error is very similar to the error you reported the first day. If it happens to every person you mail to, there is probably something wrong in kmail (at worst, a problem at your provider house: I don't believe it). If it is with only one person, that persons has problems, not you. I would setup kmail to send to localhost instead, and let postfix handle it. There is more control.

When authentication for either of those addresses is added - Get message that authentication failed, most likely the password. (password IS correct.)

Who knows? Er... I think I'll skip the rest of your message, because I think most questions are covered above. I would be repeating myself.

...

Don't touch that. To know what mails are in there, and why, use "mailq". If you need to hold, delete mails, use postsuper. If you want to have a look, use "mc" to browse files. If you want to read a mail that is there, use "postcat file".

OK, for curiosity's sake, I went and browsed those directories.Out of the hundred or more directories, there is not one mail in there. 0 -nothing - only contains one pid file of 17 bytes.

That's normal, 'mailq' showed nothing.

...

For example, "postcat -q 28D0720D17|less" displays my queued email (one) by that ID.

Couldn't do that, nothing there. (have two messages in the outbox of Kmail though)

Ok, postfix will not see them.

...

No. Unless you have configured postfix to relay to your ISP, and that is not the default config. It could be, but I doubt it.

But, when I looked at main.cf it showed relay host=sanctum.com

The space, as Patrick said. -- Cheers, Carlos Robinson

Fri, 24 Sep 2004, by usr@sanctum.com:

On Thursday 23 September 2004 20:13, Carlos E. R. wrote:

...

On Thursday 23 September 2004 07:59, Bob S wrote:

...

...

First of all at end of the last message Carlos asked about my system.

...

I was more interested in your network configuration. One computer or more,

Also I referred to the fact that looking at the /var/log/mail file a lot can be known about what is the problem/s. That is important.

OK, but that file is about 3.5 megs. About 99% of it is spamd messages on each incoming mail. Here are a few lines from Sept.22nd/23rd. ( I mostly work on this box from about 9pm until 2am US EST )....:

Sep 22 21:44:20 linux postfix/postfix-script: starting the Postfix mail system Sep 22 21:44:21 linux postfix/master[1589]: daemon started -- version 2.2-20040616 Sep 23 00:57:46 linux postfix/postfix-script: refreshing the Postfix mail system Sep 23 00:57:46 linux postfix/master[1589]: reload configuration Sep 23 00:57:46 linux postfix/postsuper[3365]: warning: bogus file name: defer/.directory

Do *not* ignore Postfix warnings about it's environment. When things are broken you can not trust Postfix (or any other program) to do its job properly. If you installed from source run 'make upgrade' to fix problems with permissions, missing files etc., otherwise run 'postfix check' and see if you can fix things by hand. Theo -- Theo v. Werkhoven Registered Linux user# 99872 http://counter.li.org ICBM 52 13 27N , 4 29 45E. + ICQ: 277217131 SUSE 9.1 + Jabber: gurp@nedlinux.nl Kernel 2.6.5 + MSN: twe-msn@ferrets4me.xs4all.nl See headers for PGP/GPG info. +

On Sunday 26 September 2004 08:35, Carlos E. R. wrote:

The Friday 2004-09-24 at 03:21 -0400, Bob S wrote:

...

...

Also I referred to the fact that looking at the /var/log/mail file a lot can be known about what is the problem/s. That is important.

Here is the last part of /var/log/mail: ( the last few lines of the last download) Sep 27 22:21:43 linux spamd[1469]: connection from localhost [127.0.0.1] at port 1373 Sep 27 22:21:43 linux spamd[26976]: info: setuid to Bob succeeded Sep 27 22:21:43 linux spamd[26976]: processing message <200409261701.11053.suse-linux-e@Trans-Star.net> for Bob:500. Sep 27 22:21:45 linux spamd[26976]:Sep 27 22:21:43 linux spamd[1469]: connection from localhost [127.0.0.1] at port clean message (0.0/4.0) for Bob:500 in 1.9 seconds, 2324 bytes. (then I disconnected and reconnected after midnight to send a reply to your message which failed - SuSE linux ....unknown...whatever.... /var/log/mail then produced this: Sep 28 01:03:25 linux postfix/pickup[17109]: 50F4A232FC: uid=0 from=<root> Sep 28 01:03:25 linux postfix/cleanup[18356]: 50F4A232FC: message-id=<4158F09B.mailLEK16Z7JD@linux.local> Sep 28 01:03:26 linux postfix/qmgr[1617]: 50F4A232FC: from=, size=8099, nrcpt=1 (queue active) Sep 28 01:03:27 linux postfix/local[18358]: 50F4A232FC: to=, orig_to=<root>, relay=local, delay=3, status=bounced (unknown user: "bob") Sep 28 01:03:27 linux postfix/cleanup[18356]: 083452333A: message-id=<20040928050327.083452333A@linux.local> Sep 28 01:03:27 linux postfix/qmgr[1617]: 083452333A: from=<>, size=9722, nrcpt=1 (queue active) Sep 28 01:03:27 linux postfix/qmgr[1617]: 50F4A232FC: removed Sep 28 01:03:27 linux postfix/local[18358]: 083452333A: to=, orig_to=, relay=local, delay=0, status=bounced (unknown user: "bob") Sep 28 01:03:27 linux postfix/qmgr[1617]: 083452333A: removed Sep 28 01:18:08 linux poll.tcpip: Starting mail and news send/fetch Sep 28 01:18:15 linux poll.tcpip: Done mail and news send/fetch ???? Makes no sense to me. Done ... tired... I quit for now .. try again tomorrow night...

...

<snip> ....

Previous to quitting:

Ah, only the log part near a problem is interesting. However, there is something funny:

...

Sep 22 21:44:20 linux postfix/postfix-script: starting the Postfix mail system Sep 22 21:44:21 linux postfix/master[1589]: daemon started -- version 2.2-20040616 Sep 23 00:57:46 linux postfix/postfix-script: refreshing the Postfix mail system Sep 23 00:57:46 linux postfix/master[1589]: reload configuration Sep 23 00:57:46 linux postfix/postsuper[3365]: warning: bogus file name: defer/.directory

Look, as root, in the directory "/var/spool/postfix/defer". There should only be directories by the names "0", "1",...,"9", "A",...,"F" (ie, 16 directories). Ummm... I only have 13. Don't have a 4, 8, or B.

If there is a ".directory" there, delete it. If it is not empty, move it temporarily somewhere else, just in case.

OK, was there and contained two lines, url properties and icon size=32. Just renamed it to old.directory.

Use a console or mc to look there, not konkeror or similar. It's a weird file to be created there.

OK, use MC since my 3.1 Norton days .....<snip a whole bunch more>....

...

...

Then, we go to Yast, and configure mail the easy way:

**** Yast ============

Start "yast", go to "Network Services/ Mail Transfer Agent".

Connection type : permanent. Enable Virus Scanning: you choose. Next

Outgoing mail server None Masquerading - empty Authentification - empty

Done ...... I assume that all of this is empty because we are going to use kmail ?? and do not have a permanent connection. Use dial-up.

...

...

Incoming mail - I'll paste it (use fixed width font to view):

............<snip a whole bunch more>....

...

...

+Incoming Mail-------------------------------+

|[ ] Accept remote SMTP connections | OK, the checkbox is unchecked... | +Downloading-----------------------------+ | There is in a dropdown menu titled "Incoming Details" which include 3 items, Downloading, Aliases, & Virtual domains. Downloading is empty - nothing there None of the Server, protocol, remote user, password, etc is there. | | |Server Protocol | | | | _____________ AUTO________ | | | |Remote user name Password | | | |_______________________ _______________ | | | |Local user | | | |____________________________________ | | | | [Details...] | | | | +----------------------------------------+ | |Forward root's mail to | |bob_________________________________________| <==== That is shown in the box above "incoming details" and is OK. |Delivery Mode | |Through procmail___________________________ | Nope - the above is not there. | [Aliases...][Virtual domains...] | As stated, they are shown in the dropdown box. Aliases has a whole bunch of

Now, I am really feeling stupid here. Following is your new table. Looks basically like the previous table but without the tiny x's So, maybe I am dumb and don't understand what I should be doing here :-( things in it (40) including "root bob" Nothing is set in Virtual Domains

...

...

+--------------------------------------------+

I hope you can see it now :-)

Hmmm.... evidently I saw it before in a slightly different format. So it was not your fault. I guess I did/not - do/not - understand what should be in there except for what I have described as "not" being there. ......Does that make sense ? I'm not sure I am on the same page with you. :-( Question: My main problem is outgoing mail. Which as above, nothing was done in Postfix because I assume we are going to let kmail handle it. For the problematic incoming section of Postfix, I assume you are being kind and helping me setup my internal mail handling. Right?

...

...

**** Some testing

Negative on the testing. Need to input the correct data first. :-(

Ok, I'll wait.

Welcome. Interesting problems are similar to a game of chess. A mind exercise. :-)

I am soooooooo.... glad that you find it challenging enough to continue with it. Bob S.

The Tuesday 2004-09-28 at 07:02 -0400, Ken Schneider wrote:

vvv

...

Sep 28 01:03:27 linux postfix/local[18358]: 083452333A: to=

name starts with lower case "b"

Just a shot in the dark.

I just tried to send to myself, putting an uppercase. I received it, so, no, that's not the problem. Mail names are case insensitive. Poor windows users, if not :-p -- Cheers, Carlos Robinson

On Tuesday 28 September 2004 15:50, Carlos E. R. wrote:

........<Bob snips a whole bunch>...........

A message that was originally sent to root, is forwarded to bob@linux.local. It is a bounce from a message sent to "bob", because there is no user by that name. That's weird.

The bounce itself (50F4A232FC) is sent.

A bit above those lines, perhaps, the explanation for the unknown user could be found. Perhaps it is searching for "bob" at some other domain; perhaps somewhere you have "bob" instead of "bob@linux.local" as return address or something.

I can't really know.

Where should "bob@linux.local" be ?? <snip> ....

...

Previous to quitting:

...

Ah, only the log part near a problem is interesting. However, there is

something funny:

...

Sep 22 21:44:20 linux postfix/postfix-script: starting the Postfix mail system Sep 22 21:44:21 linux postfix/master[1589]: daemon started -- version 2.2-20040616 Sep 23 00:57:46 linux postfix/postfix-script: refreshing the Postfix mail system Sep 23 00:57:46 linux postfix/master[1589]: reload configuration Sep 23 00:57:46 linux postfix/postsuper[3365]: warning: bogus file name: defer/.directory

Look, as root, in the directory "/var/spool/postfix/defer". There should only be directories by the names "0", "1",...,"9", "A",...,"F" (ie, 16 directories).

Ummm... I only have 13. Don't have a 4, 8, or B.

Funny. I don't know if it is really a problem, but you can create those directories using the same permissions as their "coleagues". You could also check the rest of the queues, they should have the same structure.

OK, related or not? I added the 3 missing folders to Defer & Deferred. The Directories Bounce, Active, & Incoming were correct. The Directories Corrupt, Flush, Hold, Maildrop,Saved, & Trace, are all empty. BTW even the folders with 1to 9 A to F are empty. Nothing in there.

...

...

If there is a ".directory" there, delete it. If it is not empty, move it temporarily somewhere else, just in case.

Gone !!

...

OK, was there and contained two lines, url properties and icon size=32. Just renamed it to old.directory.

...

...

...

...

Then, we go to Yast, and configure mail the easy way:

**** Yast ============

Start "yast", go to "Network Services/ Mail Transfer Agent".

Connection type : permanent. Enable Virus Scanning: you choose. Next

Outgoing mail server None Masquerading - empty Authentification - empty

Done ...... I assume that all of this is empty because we are going to use kmail ?? and do not have a permanent connection. Use dial-up.

Me too, it doesn't matter. Using what they call non permament netowrk connection has some other side effects I don't like.

OK

But I prefer you set up postfix the other way, no relay, for the moment, at least till you run the tests I sugested.

OK

...

...

...

...

+Incoming Mail-------------------------------+

|[ ] Accept remote SMTP connections |

OK, the checkbox is unchecked...

It can only be used if you have a permanent network connection and a domain name with an MX entry.

OK - explained and understood

...

...

...

...

| +Downloading-----------------------------+ |

There is in a dropdown menu titled "Incoming Details" which include 3 items, Downloading, Aliases, & Virtual domains. Downloading is empty - nothing there None of the Server, protocol, remote user, password, etc is there.

Right.

The downloading part would serve to configure /etc/fetchmail. This program is called from the script /etc/ppp/poll.tcpip - you can see when it is called at the end of your log excerpt I left above. If it is empty, it does nothing (except trigger postfix sending).

OK, I saw it and checked it, but No, not empty, bunch of commands in there. But the downloading part is empty,

The aliases is used to forward root's mail to bob.

OK - have that

The virtual domains is used to loop internally (in postfix) mail sent to yourrealaddress@somedomain.somewhere.

OK, this part confuses me. What should be there? bob@linux.local ? or my real ISP address?

...

...

...

...

| |Server Protocol | | | | _____________ AUTO________ | | | |Remote user name Password | | | |_______________________ _______________ | | | |Local user | | | |____________________________________ | | | | [Details...] | | | | +----------------------------------------+ | |Forward root's mail to | |bob_________________________________________| <====

That is shown in the box above "incoming details" and is OK.

...

...

...

|Delivery Mode | |Through procmail___________________________ |

Nope - the above is not there.

Well, it is distributed diferently, but it is there.

...

...

...

...

| [Aliases...][Virtual domains...] |

As stated, they are shown in the dropdown box. Aliases has a whole bunch of things in it (40) including "root bob"

Right.

Again, OK as stated above

...

Nothing is set in Virtual Domains

Put:

your_email@yourisp.hisdomain bob

As above - the part that confuses me. And, if I add the "bob" at the end I get a message that this is an incorrect format. ie: "rnr@sanctum.com bob"

This would avoid mail sent internally to your "from" address to go through your ISP server and back to you.

Which is what the above should do?

Postfix works almost out of the box, with a little help from yast - it should be working by now, only some glitches. Dificult to diagnose from such a distance.

I understand. Not a problem, and appreciate the help.

Don't forget doing the tests I wrote the other day :-)

OK, Bob cuts and pastes from the previous message about testing. **** Some testing

Do "sux -" on an xterm, or open a root console in kde's Konsole. Type:

nimrodel:/etc/postfix # mail cer Subject: test hola .               <==== the dot\n signals the end of the email. EOT nimrodel:/etc/postfix # nimrodel:/etc/postfix # You have new mail in /var/spool/mail/cer nimrodel:/etc/postfix #

OK, did that. (substituting bob for cer of course) Nothing - no notification - no entry in /var/spool/mail/bob

That is, send a manual email to yourself. If after a minute, when typing enter on bob's console (on any bob's console, in  fact), you see the message that you have a new email, all is well - so far :-). You may try to send another to 'root', it should also appear there.

You may - no, do - check it looking at the '/var/log/mail' file. You will see entries from several postfix's processes, and perhaps amavis and/or spamassassin. The end will be something like:

Sep 24 00:00:05 nimrodel postfix/local[9618]: 35DAA20D14: to=, orig_to=, relay=local, delay=1, status=sent (delivered to command: /usr/bin/procmail)

/var/spool/mail is empty except for a 0 byte file named "bob" No amavis, no spamassassin, nada ! Terminado de la prueba ! Por ahorita !! **** Advanced settings, postfix. **** Sending outside. **** Configuration possibilities. All on "Hold" for now" but will be filed for future reference. Carlos, You have been very helpful and I have learned a great deal from you. This is starting to get ridiculous and I would not blame you in the least for dropping this thread. Beginning to think there is some other hidden problem in KDE or Kmail 1.7 which is not apparent. Maybe I will just wait it out until 9.2 or whatever the next release is called and do an upgrade. Ummm...Could it have anything to do with ipv6 ??? I see that in my "hosts" file. Just a wild guess. Bob S.

On Tuesday, 5 October 2004 02.52, Carlos E. R. wrote:

When postfix is working, it saves a file whose name starts with B* (for example), in a directory starting with the letter B, like ./Defer/B/Bfilename

Perhaps it makes faster searching when there are a lot of messages going around, in the thousands. The trick worked in MsDos, I know, but I was not aware it also worked in *nix.

A directory is a file, rarely sorted in alphabetical order, containing your filenames and the inodes they point to. When you have all files in a single directory, the OS will have to open the directory file and read it sequentially starting from the top. On average, this means opening a file in a standard file system is an O(n) operation, where n is the number of files in the directory structure (if a single directory can be called a structure) Assuming the hash algorithm is balanced enough, doing it the postfix way means you get a 26th as many reads on average, and if you take it to the extreme (a complete tree, with one file (and its collisions) per directory) means you get O(log n) instead of O(n). When you have a huge number of files (or a slow disk system), that can make an enormous difference. It's a pretty standard algorithm for speeding up searches I'm not sure if reiserfs is affected by this, since it already stores files in balanced trees. I'd have to look that up

The 2004-10-05 at 03:34 +0200, Anders Johansson wrote:

...

Perhaps it makes faster searching when there are a lot of messages going around, in the thousands. The trick worked in MsDos, I know, but I was not aware it also worked in *nix.

A directory is a file, rarely sorted in alphabetical order, containing your filenames and the inodes they point to. When you have all files in a single directory, the OS will have to open the directory file and read it sequentially starting from the top. On average, this means opening a file in a standard file system is an O(n) operation, where n is the number of files in the directory structure (if a single directory can be called a structure)

Right. In Dos, the problem when opening large directories was that the directory itself, the index let's say, was spread over several records, and often fragmented, thus making the operation slower. Then the optimization needed was to subdivide in enough directories so that the OS only needed to read a sector to find where the file was stored (or where it started, rather).

Assuming the hash algorithm is balanced enough, doing it the postfix way means you get a 26th as many reads on average, and if you take it to the extreme (a complete tree, with one file (and its collisions) per directory) means you get O(log n) instead of O(n). When you have a huge number of files (or a slow disk system), that can make an enormous difference. It's a pretty standard algorithm for speeding up searches

I'm not sure if reiserfs is affected by this, since it already stores files in balanced trees. I'd have to look that up

According to the developers, no, it is not affected: it is designed for such a job. They say that you can implement a database using the filesystem itself as the engine, storing a record per file, or even a field per file if you want. This would further improved by plugins. I believe there is a perl database that demonstrates the concept. -- Cheers, Carlos Robinson

The 2004-10-05 at 03:29 -0400, usr@sanctum.com wrote:

Virtual Domains = rnr@sanctum.com (destination is bob@EasyStreet) ( I could not make it accept the example you gave me - Kept saying it was the wrong format)

The format was correct for entering directly into the '/etc/postfix/virtual' file, and then running 'postmap virtual'. If you enter it in Yast, it may differ. I don't have a suse 8.2 to check.

OK - Onwards & upwards - Did the "testing" procedure you gave me. Would not mail to "bob" but would mail to "bob@EasyStreet" Could not mail to "root". "root@EasyStreet" nor "EasyStreet". Following are the last few lines of the /var/log/mail file: ---------------------------------------- Oct 5 02:42:43 EasyStreet postfix/local[27471]: A6A222333F: to=, orig_to=, relay=local, delay=12, status=sent (delivered to command: /usr/bin/procmail) Oct 5 02:42:43 EasyStreet postfix/qmgr[3309]: A6A222333F: removed --------------------------------------------------------------

That entry shows that mail sent to root@EasyStreet.local was redirected to bob@EasyStreet.local - that's correct. The names that did not work will probably be because the machine name is not recognised as EasyStreet. More about that on the other thread.

I also followed Ander's advice and changed everything I could think of to lower case letters.

Safer, yes. -- Cheers, Carlos Robinson