Topic: Sending mail from Clients Problem

[Craig Main]

Hi all,

I have recently installed a server with SME Server 5.6. All seems to be
setup fine, however...  there seems to be an intermittent problem with
the mail. The clients are running Outlook (various flavors). When
checking the mail settings, outlook reports everything to be fine. When
trying to send a mail however, it fails with a smtp timeout error.

I have installed amavis-ng and clamav. The setup for that seems fine.
The error occurs whether I have the qmail-queue.amavis or the
qmail-queue in use.

Has anyone seen this before? Got any pointers?

The server is also doing file sharing and that seems to be working fine.
POPing mail from the server appears to be working fine too.

TIA
Regards
Craig

[Craig Main]

I am getting this in the Logfiles when the client tries to send mail:

Jun 18 14:11:55 hope PAM_pwdb[13854]: authentication failure; (uid=101) -> admin for pwauth service

Please can someone help?

TIA
Craig

[Howard Jacobson]

I am having the exact same problem, only I've not done anything to my server in months, and the problem just arose today.  I can ping my mail server but SMTP from Outlook 2000 times out.

HAJ

[Howard Jacobson]

One additional data point is that my smtpfront-qmail log file has entries like this for every time I try to send mail from Outlook but get a timeout error:

@400000... smtpfront-qmail[nnnn]: bytes in: 0  bytes out: 38

The "..." indicates a hexadecimal number.
The "nnnn" indicates a four digit base-10 number.

Thanks.

[Charlie Brady]

Howard Jacobson wrote:
 
> @400000... smtpfront-qmail[nnnn]: bytes in: 0  bytes out: 38

That means your (Outlook) clients connect, but then send no data (0 bytes).

Charlie

[Howard Jacobson]

>> @400000... smtpfront-qmail[nnnn]: bytes in: 0 bytes out: 38

> That means your (Outlook) clients connect, but then send no data (0 bytes).

Well, the question then is why Outlook isn't sending anything because Outlook works fine when I change the SMTP server to point to one of my other e-mail accounts.  Any ideas?  Again, this just suddenly started today.  Thanks.  HAJ

[Ben Hinton]

I'm also having the same problem. Outlook 2000 has been connecting through my E-smith server no problems up until yesterday.

Being a beginner to E-smith, I have no solutions to offer

Best wishes,

Ben

[Greg Jones]

Did anyone sort out what was causing this problem? It's now happening at our site as well. We haven't made any changes to configuration for some time.

This is all very wierd.

Thanks in advance for any help.

Greg Jones
Jacus - SoftwareWorks

[Greg Jones]

As a follow-up:

I've turrned up the timeout from 1 minute (default) to 2 minutes on my Outlook clients and they eventually do get a response from the SMTP server and manage to send mail. So, it looks like it's really a qmail problem.

Now I just need to find out why qmail is suddenly running so slowly.

Any ideas, anyone?

[Greg Jones]

Yet another follow-up

We have our system setup to use spamcop (amongst others) to check incoming connections against the spam black-lists. According to another message on this board (http://forums.contribs.org/index.php?topic=8072.msg30121#msg30121), there have been a lot of DOS attacks against these sites lately.

If you are using similar software, there may be delays in validating the originating IP address and this may be the cause of the delay in accessing the SMTP server (it can take over a minute for my server to display its welcome message).

I haven't tried turning off the spam checking to test out this theory yet, but it seems like a plausible explanation (at least to me).

Hope this helps,

Greg J.

[Brian Hand]

I have exactly the same proble, just appeared this morning out of the blue!

[Darrell Lewis]

I have a site that has just developed this problem too.

Sending email from Mozilla or Eudora - 50-60 seconds watching the sending window then - whoosh it's gone.

Anyone found a resolution to this one yet?

[Van]

This is a problem here as well -- my users are from 3 sets of public IP addresses -- we have over 500 users -- many are complaining of very slow outgoing mail & timeouts -- any way to raise priority of SMTP?

[Charlie Brady]

Van wrote:

> This is a problem here as well -- my users are from 3 sets of
> public IP addresses

Point 1: SME server isn't designed for clients sending from public IP addresses - usually clients are on a local network with private IP addresses.

Point 2: If there are delays, it's likely because reverse DNS isn't set up correctly for the sending IP addresses. Reverse DNS for those addresses is not under the control of the SME server.

Charlie

[Van]

Ah, Am I using the wrong solution? we are a small ISP -- I do have the public ranges defined as local networks.

I'll have to research about the reverse DNS

Van

[slang]

Van wrote:

> This is a problem here as well -- my users are from 3 sets of
> public IP addresses

Point 1: SME server isn't designed for clients sending from public IP addresses - usually clients are on a local network with private IP addresses.

Point 2: If there are delays, it's likely because reverse DNS isn't set up correctly for the sending IP addresses. Reverse DNS for those addresses is not under the control of the SME server.

Charlie

Hi Charlie and Co.

Interestingly, I too seem to be having this problem, ever since putting in 6.0. I'm no bunny when it comes to mail systems, but this one is starting to annoy me, hence I'm likely to start missing obvious solutions...

My SME 6.0 is on a DMZ (192.168.3.0/24), and it's authorive dns for the networks that my LAN PC's are on (192.168.2.0/24). This is of course a different network than the local eth0 interface on SME.

The LAN network is listed in SME as being a local network, an I've configured hosts for my internal PC's.

For some reason, SME just won't do a successful reverse lookup, even though it should be authoritive. dig -x 192.168.2.129 fails, although dig -x 192.168.3.1 succeeds.

My LAN PC's are on 192.168.2.0/24, and because SME can't reverse resolve these, it takes time.

I'm not familiar with the tinydns/dnscache software, so haven't made much progress yet.

The same configuration worked fine with SME 5.6 (which was based on bind.)

Anyone have any suggestions?

[cyberdude]

Hi!!

I have the same problem, but with the buield in webmail in the sme 5.6
In the log i can se a similar posts "did not recive identification string from..."

But this problem ONLY accours when i conects to the internet throug our new router with DHCP

If i put my netcable directly to our ISP's adsl modem i easily can send mail from webmail but if the router is between me and the adsl modem i get timeouts

Havent find a solution.... yet

[Van]

This problem is now front and center -- my mail server itself has no reverse dns - mail is being rejected from my users sending to other domains --

will it help to create an entry  in tinydns for a reverse record? I do not have a dns server right now

Should I go to my ISP - Sprint and have them create a reverse record?
I am a bunny on this stuff, or maybe dumber than that

Will appreciate the help

[van]

Got Sprint to create the reverse DNS -- some of the problem mail is going through now. I'm now looking to build the dream mail server -- I want:
 
1 host multiple domains and IP ranges
2 offer secure email with encryption
3 I'd like a probation / white list system that returns mail from senders that the user hasn't yet appoved until they send a form or such to verify who they are
4 I want everything else that I haven;t thought of but will realize I should have between now and now+2 yrs

and now for the howto for the complete bunny?

[smeghead]

Hi all

I recently ran into a similar problem on a client site with some systems.

The upshot of my investigation was that Nortons Antivirus was the culprit - it was messing with the mail before it even got to the SME server and causing the delays or timeouts.

As we were moving to a different A/V product we basically uninstalled it and the problems disappeared.

HTH

[cyberdude]

hi!

I have no antivirus program instaled on the clients that are unable to send mail throug webmail....

I have now 2 clients that have an ?DSL conection with their ovn router at the LAN net....., and when conecting directly throu the ISP the problem dissapere....

So the problem is in the router / firevall in the router / NAT-DHCP section in the router...

[rshiras]

Outlook status bar says receiving 1 of n messages because there are n messages in the <user>/Maildir/cur directory.  If I log into webmail, I see that I have already deleted the e-mails that these files (in the cur directory) represent.
So I think the problem is that qmail is not removing the files when it should.
Are we getting closer to a solution?

[ryan]

I am just throwing this out....MS has released several critical updates in the past few weeks.  Could these be impacting Outlook?  Many suggest the problem just started recently.

ryan

[precisenz]

I've got the same problem since we upgraded a client from 5.6 to 6.0.  They're using Outlook 2003 with all the latest updates on one machine, plus Outlook 2002 on another.

The only way they can clear their mail each day is to try it, and reboot if it doesn't work.

Software being used:
SME 6.0, Server only mode
Windows XP & 2000
Outlook 2002 & 2003

Any further ideas?

[Brenno]

I may be suffering from similar problems.

More often than not, when I try to send email from Outlook Express through my SME6.0b2 server via IMAP, Outlook cannot move the message into the sent-mail folder after it has sent (which is what I configured OE to do,) and in some cases, it doesn't appear to be sending the mail at all.

I've always had a problem moving mail from Outlook Express' local folders on the SME's IMAP folders (see my previous posts,) but now this may be affecting mail delivery.  3 test messages sent to another account where Express couldn't move the file to the sent-items folder on SME weren't delivered at all.

I can't find any errors in any of the logs on the server.  This has been going on since I set up the server less than two weeks ago, but it is getting worse each day.  I have increased the timeout limit in Outlook, but that's ineffective since the transaction is not timing out - it's simply not occuring at all.

Would it make sense to specify my ISP's SMTP server in the email settings of SME so it's used instead of the native SMTP engine?

[Brenno]

Since I was concerned that this was an Outlook express problem, I connected to my SME server at home via Evolution.  To my dismay, I had the same problem!  Then I logged in to the same account via webmail and found difficulties there.  Could this be SME?

Here's some symptoms:

In Evolution, when sending mail, it reports an internal error but indicates the mail was sent.  A copy of the mail is left in the local folder Outbox and the mail is confirmed as being delivered.  If I hit Send/Receive, a second copy of the email is delivered and the message is moved from the local folder Outbox to the sent-mail items in the IMAP folders where two copies of the same message now appear!

In Outlook Express, when sending mail, it reports an internal error.  No indication is made as to whether or not the mail was sent (sometimes it is, sometimes not.)  No copy of the email is left in a local folder, and it isn't moved to the IMAP server folders - it's essentially gone.

In WebMail, I can compose/forward a message and appear to send it fine.  But I sent three messages to a test account and only found two of them in my sent-mail folder afterwards.  (In all cases, "save a copy in 'sent mail'" was checked.)  All three emails were delivered.

Looks to me like SME is having some problems putting sent messages into the right place when it's done.  Anyone else encountering this?  Is it possibly related to other issues in this thread?  Are they symptoms of the same problem?

[Brenno]

In an effort to see if other clients could work properly, I installed Thunderbird on my laptop and used that.  It, too, has the same problem - it's sending email just fine, but every time it attempts to move the message to the sent-mail folder on the server, I get an internal message.

The only good thing about Thunderbird is that it prompts to see if you want to try again.  When you select Yes, it moves the message to the sent-mail folder without any problems.

I'm pretty surprised that nobody has responded to this thus far.  I don't understand how I could be the only one this is happening to, or who has had this happen to them in the past.

Nobody has any suggestions for a fix?  Ideas as to what to look for?

Help?

[Jon_Reynolds]

Has anybody found a solution to this problem? I am now seeing it here as well.

What I did to try and fix this was to install secure email and then reconfigured outlook to authenticate from the server. This way the server would allow a user to relay through the server. It stopped the 451 not a valid rcpthosts but now it is taking a long time to connect using telnet to the 465 port. Using telnet from one of my linux boxes it connects to the ssl port of 465 instantly. Doing the same from an M$ client it never seems to connect to the port. I am not sure if this is how windows operates when trying to connect to a secure port or not.

Is there a contrib that will allow someone to easily setup tinydns so I can use my sme 6.0.1-01 as the dns and configure reverse dns to see if that clears it up?

Some where there has to be a solution for this error.

Jon

[Brenno]

I'm at wits' end here!

I've finally tried storing the sent emails on a different folder on the IMAP server to see if that was related somehow.  Of course, I still experience exactly the same problem.  Virtually all clients tried (including webmail,) fail to move a copy of the sent message to any folder on the SME server after sending the message.

Is this a bug in SME's implementation of IMAP?  I'd assume not, or they're be more discussion on it.  Is it then specific to my installation and if so, does anyone have any suggestions as to what to look for and/or how to fix it?

I hate to consider this, but I might have to move all my clients back to POP mail.

[Jon_Reynolds]

Not sure if this will help anyone else but I fixed it on my end. I had 3 problems that I needed to solve.

My setup was DSL-->Router-->SME-6.01-01

 My users kept getting the dreaded '451' rcpt hosts error. This one was easy, we were doing DHCP from our router so qmail was thinking they were not local. I made all clients reboot and used DHCP from the SME and that solved the 451 error locally.

 I installed 'securemail' from pagefault.org and used smtp auth so that my users outside of the local network that were not getting DHCP from the SME could send through my SMTP server.

 What I was seeing was that certain emails would send but others wouldn't. The ones that would send were going outside my ISP, e.g. hotmail,yahoomail...etc, I could even receive emails from these sites as well.

 The ones that would not go out were going through my ISP, my clients were configured to use our ISP to send and our POP3 server to receive. Yet even when we sent through webmail the same problems would occur. We set everyone up to use SMTP Auth and the problem remained.

I would send an email and it would simply disappear. It would leave my network and what happened to it I had no idea.  

I re-installed the SME and at the point were it asked about my DNS server I put my ISP's DNS address and all of a sudden everything started working. The mails that had disappeared were bouncing back to my ISP and sitting in their queue.

Maybe this doesn't help anybody else but maybe it will,

Jon

[Wooderson]

I think I'm seeing the same issue with a client here too.

Contribs.org 6.0 release.
Securemail conrib, clam av and spam assassin are installed.
ALL the clients with outlook 2000 work fine, however the clients with Outlook XP will time out every time.

Tomorrow I'm going to run a "netsh int reset c:\resetlog.txt" on one of the problem clients and then pull Outlook XP and replace it with Outlook 2000 (since all the other OL2000 clients work fine) if that doesn't do it.

I haven't checked the logs to see if the error is exactly the same, but I will.

[Brenno]

I've done more testing in an effort to fix this, and I will continue to post progress on this forum in case it's of benefit to anyone else (which is what this forum is for, IMHO.)

So, I brought the old 5.6 server online again and began using it for testing.  Immediately, I noticed that I was not having any of the problems using this server as IMAP as I am with the 6.0b3 machine!

Therefore, I must conclude that this is either:

a) a problem with 6.0b3 itself not found in 5.6, or,
b) a problem with my installation/configuration.

Unfortunately, determining which option it is will be difficult as it will require significant work to prove either case.

[Jon_Reynolds]

Hey Brenno,

Is there a reason why you are using the 6.0b3 instaed of 6.0.1-01?

Jon

[Brenno]

Jon,

Good question.  I'm using 6.0b3 as it's the last version I downloaded and burned as an ISO for installation.

Do you feel that 6.01-01 would be more reliable?  I'm open to suggestion at this point.

My current plan of action is to move all email to the 5.6 server and go live with it until I can figure out the best option with the 6.0b3 machine.  I am considering installing 5.6 on it, too, and going from there - but I am starting carte blanche.  And I'm going to test the new machine to death before I implement it - that was likely my problem with the last time.  Had I been more patient, I may have done things differently and avoided this.

I gave some thought to the possibility that the problem has something to do with the fact that I have 4 drives in the 6.0b3 machine.  2 are as RAID1, the other two I added as per a howto listed within the forums here, one drive mounted at /home/e-smith/files/users the other at /home/e-smith/files/ibays.  This provides maximum space for data as I couldn't get it to install as a RAID5 using the hardware in the machine. The install saw right through it and only installed on one drive.  The catch with this setup is that quotas just aren't working properly, but that didn't seem significant at the time.  I considered simply reversing the procedure in the howto, but didn't want to do something so big while the server was live.  Maybe this has some bearing on it?

[Jon_Reynolds]

Well the latest one is 6.0.1-01 from the download page, so that is what I would use. The b3 means obviously beta and I wouldn't want to use a beta version in production.

I went back and re-read your posts, is mail working to some locations and not to others? What do the mailer-daemon messages tell you? Can you see your message leave your network through the logs?

SSH into your SME box and do a 'tail -f -n 30 /var/log/smtp-qmail/current' and watch as each message. Then, while monitoring that send out 3 emails one to someone internally, externally and to your ISP where you have an account, which messages show up?

Did they all leave your server and only a few show up at their intended destinations?

Is this the problem you are having or is it just trying to get your IMAP to sort messages correctly?

Jon

[Brenno]

Jon,

I guess it boils down to getting IMAP to sort messages correctly.  Actual delivery of mail to remote/local recipients was never really an issue; neither was receipt of mail from remote/local senders.  I've used two 6.0b3 machines as POP servers for the last year and half or so (since that version came out,) and have never had a problem.  I haven't tried IMAP on the other 6.0b3 server yet, though.

The next question is: what is easier - migrating the data from an existing 5.6 to a fresh 5.6 and then upgrading to 6.01-01, or migrating the data from an existing 5.6 to a freshly 6.01-01?

The answer to that will govern my next move, really.  I don't want to continue using the existing 5.6 for long because I don't trust the hardware - that's why I bought the new hardware!

[Jon_Reynolds]

Hmmm, that is the question. I have never had to upgrade from 5.6 because when I found SME it was in the progress of moving to a new version so I waited it out.

Whatever you do, make sure you have complete backups to restore from and that they work. If it were me I would setup my new server how I want it then migrate everything over. I wouldn't trust a script to migrate it over because then if something is not right I would always doubt the script and maybe miss the real problem.

What would probably be the best way is to either clone or configure another 5.6 that exactly resembles your current box. Then throw the 6.0.1-01 cd in and upgrade the 5.6 and see if anything breaks, if not, then thats the way to go.

Jon

[precisenz]

I dont think its specifically an IMAP problem as my clients are using POP with this issue.  Oddly enough if the client machine is rebooted 2-3 times it works fine without problem for the rest of the day.  Only seems to affect users that take their machines off the network overnight (Laptop users).

Since so many seem to have this problem and to begin reducing this problem down, can ppl post what they are trying so others can test it on their systems too so we all don't just hit our heads against a brick wall trying the same ol' same ol'?!

[tpearsall]

Any updates on this issue?  I have timeout errors on secure smtp connections periodically with the same 2 users out of about 12.  Inside the network the secure mail works fine, from the outside it timeouts on occassion.

I thought it may have been an MTU issue and tried lowering the MTU to 1300 on one of the laptops having trouble, but to no avail.  

Config:
- SME 6.0.1
- Secure mail smtps/pops/ssmtp
- spam assasin
- avmailgate

Thanks,
Todd

[Brenno]

As a really late followup to my initial series of posts, I have resolved all the issues I was having with IMAP and 6.0b3 by upgrading the machine in question to 6.0 final.  It's working beautifully now.

To migrate between the old and new server, I first upgraded the old server to 6.0 to ensure the file structures were the same, and I recreated the user accounts on the new machine manually.  Then I just used rsync to copy all of their content (email, files, etc,) from the old machine to the new machine.

Lastly, I shut down the old machine and reset the IP of the new one to match that of the old one.  (This meant I wouldn't have to reconfigure the external firewall's port fowarding rules or the workstation's email client settings.)

I guess it was just a flaw in 6.0b3's packages.