Koozali.org: home of the SME Server
Legacy Forums => Experienced User Forum => Topic started by: Van on December 15, 2003, 02:00:30 AM
-
I am getting outgoing failures for all outgoing mail after upgrading to 6.0b3
This is the error
The connection to the server has failed. Account: 'mail.froglevel.net', Server: 'mail.froglevel.net', Protocol: SMTP, Port: 25, Secure(SSL): No, Socket Error: 10061, Error Number: 0x800CCC0E
I've tried using authentication -- and I've tried different names for the smtp server
anyone have an answer?
-
I'm learning that part of my problem here may be that my mail users are not behind the server -- some are on the same public IP range as the server. others are on two other public ranges -- this is a an ISP mail server -- I have added all of these ranges to the "local networks"
-
Hate to keep adding to my own thread -- I created a new box from a tape backup --
I get the same errors and even in webmail I see
"There was an error sending your message: unable to connect to smtp server localhost:25"
-
tried port forwarding port 25 -- no help there --
-
Time to chuck in my 2p worth.
The error you reported, is that what you see on your MS clients by any chance?
This may/not help...
http://support.microsoft.com/default.aspx?scid=kb;en-us;302339
-
Also what happens when you telnet to port 25 on your server from your remote clients?
-
http://www.lifewithqmail.org/lwq.html#relaying
This may also be of some assitance, but I have no idea how this interacts with e-smith.
PeterG.
-
Peter,
When I telnet to x.x.x.x:25 the reply is -- could not open connection to the host,, on port 23: Connect failed.
In the reading I did last night I decided that my problem was that ports were closed on the machine. I have tried to go back to 5.6, I will try this morning to go to 6.0 final
It occurred to me that part of the puzzle is that while these servers can be seen from the internet, when I run the console line to test Connection to the internet - they fail. I have a WS backup and a tape backup -- unfortunately both of those are made from the original server after it was upgraded to 6.0b3 -- I also have an older WS backup --
My plan this morning is to install 6.0 final on a new computer, then restore files from tape -- I'm grasping for straws out of ignorance -- this server has about 1000 users -- the business users will be upset this morning.
-
It looks like that telnet test failed as it the error reported is talking about port 23 (the standard telnet port), try this instead
telnet x.x.x.x 25
This should give you a line of info, but I have not found a way of exiting this session properly (had to crash the window instead).
The test that you have performed to test connection to the internet, is that the one in the character based screen, that is also used in the initial setup?
You say that test fails, but you can still ping the external IP from the internet? This seems very strange, are you behind some kind of router with an external address range?
PeterG.
-
Router setup is plain vanilla -- other IP ranges from remote public networks that are added to the "local networks" list can even run the administration tool, or use webmail -- I left last night at least feeeling that the users would receive mail OK -- That was only because I received mail -- in looking at that mail it was just failure notices.
You are right about telnet -- but it then says unable to open port 25
It is the character based menu that reports unable to connect to the internet
In reading a thread about upgrading 4-6 threads before this one -- could I have a bad
config.db?
-
ok.
assumptions.
your users can connect to the servers external IP and use webmail
You need to check -
In the e-mail setting tab in server manager what is the retrieval mode?
In a shell can you perform the following -
What is the output of the ifconfig shell command?
can you ping external ip address from the server?
can you traceroute to the same ip address?
ps -ef | grep smtp
Please post the results of these questions.
PeterG
-
> In the e-mail setting tab in server manager what is the
> retrieval mode?
SMTP -- access from Public Internet
> In a shell can you perform the following -
> What is the output of the ifconfig shell command?
both eth0 and eth1 are up
eth1 is the public as is normal setting
loopback is running
> can you ping external ip address from the server?
yes
> can you traceroute to the same ip address?
yes
> ps -ef | grep smtp
root 376 340 0 Dec14 ? 00:00:00 supervise smtpfront-qmail
>
I also got the email this morning as described here:
http://forums.contribs.org/index.php?topic=8911.msg33578#msg33578
I've serached on this topic -- but it isn't helping yet
-
Can you try telneting from the server to itself on port 25, try on both its internal and external ip addresses.
I belive I have your IP address already from your e-mail address domain name and I have sucessfully got a login screen for web-mail and it responds correctly to pings. I am puzzeled as to why the smtp demon is not responding but the answer to the tasks above should help.
PeterG.
-
connection refused to both
The version of sme is "Unknown" --- I am upgrading one of these servers to 6.0 final to see if that helps -- thinking about the zcat error -- I may have caused from some time ago that by deleting man pages to give my earlier server room -- it paniced after running out of HD -- I have run a long time since though with no prob
thanks
-
Of course the obvious question is what is in the qmail logs in /var/log/qmail ?
8-)
PeterG.
-
notmuchdotcom
current shows 8 successful deliveries to me and my aliases
sample:
staus: local 2/10 remote 0/20
delivery 7:sucess: forward: _qp_11724/did_1+0+0/
status local 1/10 remote 0/20
lock and state are empty
the upgrade to version 6 final -- still shows (unknown version) at the top of the management console -- I'm about to check to see if it can handle mail
-
from another post on this forum I tried:
/sbin/e-smith/signal-event email-update
followed by:
/sbin/e-smith/signal-event reboot
it did not work
-
I don't know Van, I am very puzzled as to what could be wrong. If the mailfront service is running then it should accept connections. Your server can ping out to other external IP addresses, and other IP's can ping you. Your remote users can use webmail. But new connections to your mail server are timing out rather than being rejected outright.
Is your server able to send mails correctly from its own webmail?
What could be blocking access to the mailfron-smtp demon!!!
I am curious as to why the server is describing itself as unknown version too.
The quickest results may be to build a fresh 6.0 server and restore the data from tape over that.
But I am grasping at straws now. I thought other people may have chipped in by now...
Good luck and if you do find out what the problem was, please post back here.
PeterG.
-
Are you running any sort of spam or virus checker btw?
-
Perhaps there is a problem when upgrading the mailfront-smtp process on some installations as a last resort try the latest version of mailfront?
On my 6.0 server (upgraded from 6.0b3 with all updates to 6.0) I am running..
[root@pogo# rpm -q mailfront
mailfront-0.81-1
PeterG
-
I've tried several fresh installs of 6.0b3 overlaid with the backup -- I have tried upgrading one of those with v 6 final. I now have a 6 final that is virgin -- all I care about rescuing from the old copy is qmail, ibays and all dont matter -- It seems that the problems may relate to TCP/IP and ports and such beyond qmail itself
We've been using esmith since 3.0 / maybe 2.x -- If there's a way to just copy & create users and their accounts, I'd settle for that now, and just dump the mail -- I'm going to read that book you sent me the link to to see if there's a way to do this
I have called a Mitel reseller to try to purchase this product last week, before this fiasco -- he did not know about it of course, but plans to call me with a quote. I am ready to buy commercial software -- if I can move users over
-
What version of e-smith were you upgrading from?
I belive that mailfront was introduced at version 5.5 so may be if yours is before that then there might be a problem with that section of the upgrade script
If you were running 5.5 it might be worth going to 5.6 first and then to 6.0.
Just done a search for server migration and found this at the end of one link, it look like it might help....
http://tehvand.com/tehvand/index.cgi?opt=projex&project=rsyncmigrate
PeterG.
-
whups --
it was 5.1.2 -- had no plans to touch that server until thing went wrong -- I then upgraded it directly to 6.0b3
van
-
Ah.
right then. I think it might be an idea to get the iso's from the mirrors for 5.5, 5.6 and 6.0
Restore the 5.1.2 server and then go through each of the upgrades of the different versions above, with a little bit of luck this might work...
PeterG.
-
I've bought Guardian Digital's engarde mail suite -- The guys there think I can move the /etc/passwd file and the user directories, wave a wand and fix this, we'll see --
I'll also try the stepwise upgrade path on one of the other servers -- just trying to get fixed fast -- users are restless
-
PeterG wrote:
> What version of e-smith were you upgrading from?
>
>
> I belive that mailfront was introduced at version 5.5 so may
> be if yours is before that then there might be a problem with
> that section of the upgrade script
An excellent guess, sir!
> If you were running 5.5 it might be worth going to 5.6 first
> and then to 6.0.
I think the problem will only occur with upgrades from versions earlier than 5.5.
This should fix it:
/sbin/e-smith/config set smtpfront-qmail service status enabled access public
/sbin/e-smith/config delete smtpfwdd
/sbin/e-smith/signal-event post-upgrade
/sbin/e-smith/signal-event reboot
If this fixes the problem, would you mind posting a link over in the thread in the bugs forum(*)?
Charlie
(*) which is where this problem should have been reported.
Charlie
-
Charlie Cool!
Yes this worked -- outgoing SMTP is sluggish and bothers me a bit some clients are timing out -- the banner onthe character based menu now says SME server 6.0b3 instead of unknown version, food tastes better and I'm ready for what tomorrow has to offer (I hope)
I'll post the report on the bugfix
Van
-
Nice one Charlie
I was wondering if you might chime in at some point.
Glad Van got it sorted in the end without having to resort to 'the other lot'
PeterG.
-
Off topic Peter but shouldnt you be sleeping -- Thanks for your help today, and tommorrow.
Van
-
No Worries Van, Just thought I would check in before crashing for the night.
hope it works out ok.
PeterG.
-
Problems persist, users are complaining about not receving mail and outgoing mail is also slow -- a result of a poor transfer? -- box is a Poweredge 2.6 GHz with 160 Gb mirrored drives -- RAM is only 128 Mb -- got more RAM ordered -- Thoughts? logs to look at?
-
Try using the top command to see what processes are running and what the average queue wait is like.
Could also use vmstat.
If in doubt after that copy and paste the screens here for us to have a look.
What does the qmail log say?
PeterG.
p.s. memory requirement for 6.0 is 128mb so anything more will be a good improvement.
-
top reports 98% idle --
Problem is clearer now -- we are not receiving mail from the outside -- Any mail from a server user can reach another server user, but if that user tries to email himself -- say from a Yahoo account -- it does not come in -- I did a test at dnsreport.com -- it tells me that I have no reverse DNS -- this would have more effect on outgoing mail though , riight? This is a new problem with the server upgrade. The MX records at sprint look fine mail.froglevel.net is in the local names
qmail logs seem ok -- upgraded from 6.03b to 6.0 final with a ws backup and restore to a new computer -- Charlie Brady's fix made a lot of things work that were broken -- wonder if there is something else along those lines
-
I found it -- stoopid fiddling with port forwarding about 6 or 7 rebuilds/restores ago