Problem with SMTP...

system · January 16, 2006, 8:40pm

The feature is now available inthe 2.1.2 version of InterWorx.

To enalbe it

[LIST=1]

Log into NodeWOrx

Click on System Services

Click on Mail Server

Ckick on MTA Settings

Under Alternate Inbound SMTP Server, ckick Start [/LIST] The default port is 587, but you can change it if you want.

R-n-R · January 16, 2006, 9:51pm

OK I am trying to use Port 587 with Outlook Express, I did the following
First I did what IWorx-Tim said to do above in his post, port 587 on the server is showing to be active, running.

Then I opened up Outlook Express Accounts > selected properties then on the advanced Tab change Outgoing SMTP to port 587 clicked on Apply and then OK and tried to send email and got this error:

The connection to the server has failed. Account: ‘XXXXXXX’, Server: ‘mail.XXXXXXXX.com’, Protocol: SMTP, Port: 587, Secure(SSL): No, Socket Error: 10060, Error Number: 0x800CCC0E

IS there something else I need to do?

IWorx-Chris · January 16, 2006, 9:58pm

Make sure if you’re using a firewall and it’s on that port 587 / tcp is open.

Chris

R-n-R · January 17, 2006, 5:17am

Ah, yes thanks Chris

system · January 17, 2006, 9:24am

Chris, I added this port to the firewall, but for some reason the service is listed as “submission”, is there anyway to set the service name to smtp?

IWorx-Paul · January 17, 2006, 9:39am

“Submission” is actually technically the correct term for the 587 smtp port service. It’s meant for “submission” of e-mail from the end user, as opposed to communication between mail servers, which is what port 25 is for (in theory).

See: http://www.google.com/search?hl=en&lr=&safe=off&c2coff=1&q=smtp+submission+587

If you really wanted to change the name, you could probably edit the /etc/services file. Just be careful

Paul

system · January 17, 2006, 10:18am

Thanks, i’m actually going to use 1025 for it, I also have some other ports open for some other servers that I’d like to be able to have a service name associated with them in the firewall…

system · January 17, 2006, 7:26pm

Whoops, forgot to mention that . . . my bad

jimp · January 19, 2006, 9:58am

Sorry, I was actually wondering if anyone has made any progress on the original purpose of this thread (SMTP connection issues). See post #7.

R-n-R · January 19, 2006, 12:23pm

jimp I was having the same trouble from time to time. And I activated port 587 in the latest version of Interworx (2.1.2). I then set up my email manager to use SMTP PORT 587 and its been working perfect ever since. With the exception of a small problem at first with this, but it was easly fixed by opening PORT 587 in my servers firewall. Other than that works fine, solved my problem,…so far.

IWorx-Paul · January 19, 2006, 10:26pm

jimp:

I am going to jump in on this one. For approximately two months now, we have been having the same problem. POP3 and IMAP work perfectly, but SMTP connections may or may not work. I thought it was a problem with Mozilla Thunderbird, because everyone in our office uses it and no one else was complaining, but an Apple Mail user reported the problem last Friday.

The problem is difficult to reproduce, but it is effortless to cause. You simply…
[LIST=1]

Compose an email
Click "Send"
1 out of ~15 emails will block with an error message (attached to this post)
Trying again (immediately) will succeed without a problem[/LIST]
My personal testing found that SMTP w/o TLS almost never produces the error.

More evidence

We have written a photography sales system (Mozart Photo System) that sends emails to customers who have requested a reminder before photos are taken offline. I noticed that (rough guess) 10-20% of a store owner’s nightly emails would fail, although the script (called by cron) is only trying to connect to the local SMTP server. (I think that rules out the possibility of faulty data center connection.)

While trying to make Mozart’s emailer more resilient, I found some interesting properties surrounding the problem. Note that the problem can be reproduced by “hammering” the server, but I’m not talking about “attacking” the server. I just mean a script that needs to send 10+ emails in a row is very likely to surface the bug.

My test setup was trying to send an email 10 times to the local server and local email address. When the script failed to send an email, the script will enter to a recovery loop that would try 2 more times. Here is what I discovered:
[LIST=1]

Trying 3 times would fail all 3 times.
Trying 3 times, and sleeping 10s between each attempt, would fail all 3 times.
Trying 3 times, but first establishing a new connection, recovered everytime on the 2nd attempt. NO sleep was required. [/LIST]
So Mozart’s emailer is now working fine, but the mail server’s strange behavior is very similar to Mozilla Thunderbird’s. Thunderbird would fail once, but try again (make a fresh connection), and the mail would deliver just fine.

I am fairly sure this problem is very recent (within 3 months). Would an upgrade or downgrade of the mail server, and/or its dependencies be a good idea? It is possible the mail server’s memory limits are being hit?

A lot of different topics ended up being discussed in this thread, so I quoted jimp’s post to make it clear what I’m responding to here - and that is, the “intermittent smtp connection failing” problem.

After quite a bit of investigation, it looks like jimp’s final suggestion is the problem. The smtp server’s memory limit appears to be getting it “sometimes”. It isn’t exactly clear what’s causes it to happen, the memory limit was indeed getting hit occasionally, causing smtp connections to fail.

For anyone having this problem, specificially anyone that ever received the message
“421 unable to read controls (#4.3.0)”
when trying to send an e-mail message, here’s what you can do to fix it.

As root on your server, edit the /service/smtp/run
change the line


exec /usr/bin/softlimit -m 6400000 \
to
exec /usr/bin/softlimit -m 12800000 \

effectively doubling the memory limit
3) restart the smtp server: svc -t /service/smtp

Paul

jimp · January 21, 2006, 8:34pm

I am very curious to hear what you did to pinpoint the solution. My entire office, and one annoyed customer who eventually started using their ISP’s SMTP server instead, will all be very happy to hear the good news.

A while back, I also had to increase our IMAP service’s memory limit from 6400000 to 10200000, to correct a similar annoying error message about the IMAP connection being reset. I think a member of the InterWorx team helped me with that memory limit problem, too. No one in our office uses POP3. Could it be limiting out, like IMAP and SMTP were? (It is capped at the default 6400000.) Many customers just accept error messages without ever telling us, so I admit it could be happening and we would never know it.

system · January 23, 2006, 6:39am

I’ve been getting the same error once in a while with my server. What I dont understand is why it is now using more memory than it was using 2 months ago, since that is about the time the error started.