sendmail broken

[tommythejoat]

This is not really a vBulletin problem, but perhaps someone else has run into it.

I moved our vBulletin to a new machine. It is a colocation server running Centos/Cpanel/WHM. I just copied everything over there as it was on the previous machine and repointed the domain to the new machine by editing the ip address.

Everything seems to be working, but it is possible that some of the file permissions may not be correct. I did have to edit the permissions on the attachment tree.

The big problem is that sendmail is not working at all.

Since we are using email verification in registration and many users use email notifications on new threads, posts, etc. none of those features are working.

I can't even get the mail routing to work from the cpanel. Is there a good source of information on how to manage this stuff? We are not planning to use email services in any general way on this machine, so we did not do much to set up the email.

Can anyone give me a pointer on what needs to be checked and done?

[snakes1100]

You should be checking the exim_mainlog in /var/log and sending out a test mail while tailing it if you have ssh access to the server and find out the error code.

tail -f /var/log/exim_mainlog | grep "putemailherebetweenthequotes"

As you dont provide a domain, i can not check your domain for RFC failures.

[Angel-Wings]

tail -f /var/log/exim_mainlog | grep "putemailherebetweenthequotes"

Hmm - wouldn't this be the Exim log and not the sendmail one ?

/var/log/mail - maillog - mail.err - however it is called should be the file.

And is sendmail set setuid ?

ls -la `which sendmail`

Should show. Usually this setuid flag is removed on a simple copy process for a good reason ;)

chmod +s `which sendmail`

Sets this flag - of course if you really run Sendmail and not use another MTA like Postfix / Exim / Qmail.

And finally check if all Groups / Users are present with:

getent passwd
getent group

Just in case something is missing there.

Simple check if everything is working can be done while tail'ing the mail log:

tail -f /var/log/mail.err (if that's the file)

And on another console mail'ing yourself:

mail -s Test myownmailaddress@anotherdomain.com

Then some text, a Ctrl+D ends the editor and sends out the mail. This should give you some more info what could be wrong.

[snakes1100]

Hmm - wouldn't this be the Exim log and not the sendmail one ?

/var/log/mail - maillog - mail.err - however it is called should be the file.

And is sendmail set setuid ?

ls -la `which sendmail`

Should show. Usually this setuid flag is removed on a simple copy process for a good reason ;)

chmod +s `which sendmail`

Sets this flag - of course if you really run Sendmail and not use another MTA like Postfix / Exim / Qmail.

And finally check if all Groups / Users are present with:

getent passwd
getent group

Just in case something is missing there.

Simple check if everything is working can be done while tail'ing the mail log:

tail -f /var/log/mail.err (if that's the file)

And on another console mail'ing yourself:

mail -s Test myownmailaddress@anotherdomain.com

Then some text, a Ctrl+D ends the editor and sends out the mail. This should give you some more info what could be wrong.

Ok seeing as you know everything, what email program does a cpanel server use, i'll give you a hint its not sendmail.

As stated in my first post, tail the exim_mainlog, you will find nothing in the maillog file.

[tommythejoat]

I will try the suggestions so far.

I have no idea at the moment how to generate an email on this new server. In addition, it is not my email choice, but whichever one vBulletin is using. We have no intention of using this machine for serving mail.

Our domain is nawcc.org, however the vBulletin site is mb.nawcc.org. The mx record for nawcc.org points to an exchange server. All I want from the mail services on the whm/cpanel machine is to send out the mail generated by vBulletin and other applications. It would be nice to support the email integration mod, but we have another facility that performs a very similar mail reflector service, so I could do without that.

I have full root access to this machine, but I did not install the whm/cpanel. I did create the mb.nawcc.org domain on the machine and copied the vBulletin to it. I was a PDP-8/PDP-11/VAX developer for Digital Equipment years ago but I am very rusty with unix/linux. I spend a lot of time with --help.

There is currently no mx entry to mb.nawcc.org, only for nawcc.org.

--------------- Added 1255702892 at 1255702892 ---------------

The tail of exim_mai.nlog is shown below. There are lots of entries and several .gz files also, so it looks like vBulletin is trying to send out mails, but is not succeeding because the network does not believe it is authorized.

I would guess this is related to the missing mx record.

This was working just fine on our dedicated server at WestHost where they set up the server for us. We are mostly on our own with the colocation server.:confused:


2009-10-16 10:05:20 1MynQj-0001st-Nc SMTP error from remote mail server after end of data: host b.mx.mail.yahoo.com [66.196.82.7]: 451 Message temporarily deferred - [160]
2009-10-16 10:05:20 1MynQj-0001st-Nc == xxxxxxxxxxx@xxxx R=lookuphost T=remote_smtp defer (-46): SMTP error from remote mail server after end of data: host b.mx.mail.yahoo.com [66.196.82.7]: 451 Message temporarily deferred - [160]
2009-10-16 10:05:20 1MynQj-0001t7-Rw ** xxxxxxxxxx@xxxxx R=lookuphost T=remote_smtp: SMTP error from remote mail server after MAIL FROM:<nobody@centos-5.3> SIZE=2445: host mx2.comcast.net [76.96.30.116]: 550 5.1.0 Invalid sender domain
2009-10-16 10:05:20 1MynQm-0001uI-QM <= <> R=1MynQj-0001t7-Rw U=mailnull P=local S=2314
2009-10-16 10:05:20 1MynQj-0001t7-Rw Completed
2009-10-16 10:05:20 1MynQm-0001uI-QM ** nobody@centos-5.3 R=fail_remote_domains: The mail server could not deliver mail to nobody@centos-5.3. The account or domain may not exist, they may be blacklisted, or missing the proper dns entries.
2009-10-16 10:05:20 1MynQm-0001uI-QM Frozen (delivery error message)
2009-10-16 10:05:21 1MynQj-0001t2-Q3 SMTP error from remote mail server after end of data: host mx1.rog.mail.yahoo.com [69.147.95.19]: 451 Message temporarily deferred - [160]


--------------- Added 1255703505 at 1255703505 ---------------

Here is the tail from the maillog file.


Oct 16 10:24:24 centos-5 spamd[32345]: auto-whitelist: open of auto-whitelist file failed: locker: safe_lock: cannot create tmp lockfile /.spamassassin/auto-whitelist.lock.centos-5.3.32345 for /.spamassassin/auto-whitelist.lock: Permission denied
Oct 16 10:24:24 centos-5 spamd[32345]: spamd: identified spam (1000.0/5.0) for root:99 in 0.0 seconds, 834 bytes.
Oct 16 10:24:24 centos-5 spamd[32345]: spamd: result: Y 999 - GTUBE,NO_RECEIVED,NO_RELAYS scantime=0.0,size=834,user=root,uid=99,required_sc ore=5.0,rhost=localhost,raddr=127.0.0.1,rport=4293 0,mid=<GTUBE1.1010101@example.net>,autolearn=no,shortcircuit=no
Oct 16 10:24:24 centos-5 spamd[32324]: prefork: child states: I
Oct 16 10:24:24 centos-5 pop3d: Connection, ip=[::ffff:127.0.0.1]
Oct 16 10:24:24 centos-5 pop3d: LOGOUT, ip=[::ffff:127.0.0.1]
Oct 16 10:24:24 centos-5 pop3d: Disconnected, ip=[::ffff:127.0.0.1]
Oct 16 10:24:24 centos-5 imapd: Connection, ip=[::ffff:127.0.0.1]
Oct 16 10:24:35 centos-5 imapd: LOGIN, user=cpanel@localhost, ip=[::ffff:127.0.0.1], port=[50472], protocol=IMAP
Oct 16 10:24:35 centos-5 imapd: LOGOUT, user=cpanel@localhost, ip=[::ffff:127.0.0.1], headers=0, body=0, rcvd=12, sent=86, time=0

[snakes1100]

Remove the ppls emails that you posted as well.

It appears you have issues on this server in terms of setup, unfortunately, there is not much we can do from here as you failed again to supply us with a valid domain to check for RFC issues.

Your domain is not setup correctly on the server, plus there might be basic setup issues like servername being incorrect, not much we can do from here, you should have your server admin correct these issues for you.

Once you fix these setup issues, yahoo should automagically un-defer your domain, but you might need to go file a form to correct it.

The mx record will not cause these issues as the sub-domain follows the main domain.

[tommythejoat]

Remove the ppls emails that you posted as well.
Do you mean the log entries from my previous post? I don't understand ppls emails.

It appears you have issues on this server in terms of setup, unfortunately, there is not much we can do from here as you failed again to supply us with a valid domain to check for RFC issues.
I don't understand. Is http://mb.nawcc.org not an adequate domain name?

Your domain is not setup correctly on the server, plus there might be basic setup issues like servername being incorrect, not much we can do from here, you should have your server admin correct these issues for you.
Sadly, I am the server admin along with an even greener employee of our organization. WHM does complain about the name of the server when we log on. I have asked my friend to look into that. The server itself is at nawcc-host.net, but the name locally appears to be centos-5.

Once you fix these setup issues, yahoo should automagically un-defer your domain, but you might need to go file a form to correct it.
I hope so, it is really inconvenient to have vBulletin unable to generate the internal emails. It means that I have to manually register all new users.

The mx record will not cause these issues as the sub-domain follows the main domain.
I don't quite understand this remark. The mail server for nawcc.org is an exchange server on a different machine than the message board. Most of our web services are on this machine with mb.nawcc.org, but mail and business operations are on windows systems with separate machines for our database and files, mail, and .net services. These are all inside the firewall.

Our "outside" sites are nawcc.org, mb.nawcc.org, nawcc-index.net and horology.edu. We also have community.nawcc.org which is served by an asp. Those sites which are subdomains of nawcc.org are to be included in the Single Sign On environment with net.nawcc.org as the authentication server (behind the firewall).

--------------- Added 1255706561 at 1255706561 ---------------

Ah yes. ppls = people's. Done

[snakes1100]

I'll chk the domain for you.

Do you mean the log entries from my previous post? I don't understand ppls emails.

Yes, ppls = peoples

I don't understand. Is http://mb.nawcc.org not an adequate domain name?

Sorry, long morning, missed the domain.

Sadly, I am the server admin along with an even greener employee of our organization. WHM does complain about the name of the server when we log on. I have asked my friend to look into that. The server itself is at nawcc-host.net, but the name locally appears to be centos-5.

Yes, it appears you have numerous issues with the servername and with your dns entries as well, why whm is complaining.

I hope so, it is really inconvenient to have vBulletin unable to generate the internal emails. It means that I have to manually register all new users.


All this is easily fixable, nothing major.

I don't quite understand this remark. The mail server for nawcc.org is an exchange server on a different machine than the message board. Most of our web services are on this machine with mb.nawcc.org, but mail and business operations are on windows systems with separate machines for our database and files, mail, and .net services. These are all inside the firewall.

Our "outside" sites are nawcc.org, mb.nawcc.org, nawcc-index.net and horology.edu. We also have community.nawcc.org which is served by an asp. Those sites which are subdomains of nawcc.org are to be included in the Single Sign On environment with net.nawcc.org as the authentication server (behind the firewall).

The exchange server should have nothing to do with this server, so it shouldn't apply unless you routing the inbound emails thru the exchange box for some odd reason, this still doesn't pertain to the out bound email issues on the whm server.

--------------- Added 1255706561 at 1255706561 ---------------

Ah yes. ppls = people's. Done

[tommythejoat]

Our authoritative name servers are at Bulk Register. The name serving on the co-location site is intended to be "inside the firewall" type name management.

If we need additional ip addresses to sort things out, those are available, but I don't have the documentation on how to set up the virtual nic's. I would need to get that or have my associate do it.

Thanks for the help. I think I will eventually understand this, but I had not intended to learn this much about these systems this late in my career.

[snakes1100]

Yes, but when WHM sets up a site, it add's its own dns records on the server.

Additional ip's are not going to correct the issues, this is just a simple matter of correcting a few things on your server, proper hostname etc, also appears that i didnt notice that spamass is having issues as well, that shouldnt bother the out bound mail tho, but still needs to be fixed or turned off if there is no inbound mail for the domains on the server.

No problem on the help, your welcome.

[tommythejoat]

I may want to use the email integration mod for 3.7 if I can. In that case I will need the inbound mail. I set up mail accounts for the forums that will be using it and it is turned on right now but I don't think there is any traffic.

[snakes1100]

I would doubt anything will work at this point until you fix all the issues, spamass will most likely stop anything from coming in as its also broke fro mthe msg its leving in the logs.

[tommythejoat]

I may be getting out on thin ice, but I decided to do some "obvious" things. I changed the host name to host.nawcc.org from centos-5.3 and I created an A record for it locally on the whm machine.

We are in the process of moving our home site nawcc.org from WestHost to the new server, but have not yet finished that. Currently our DNS entries for nawcc.org at NSx.name-services.com (BulkRegister) differ from those on the whm system.

The entries on the whm system for mb.nawcc.org point to the NSx.name-services.com. There are 4 NS entries.

The entries on the whm system for nawcc.org are NS1 and NS2.nawcc-host.net (i.e. the local machine's name servers). I think this is necessary while we are building it from the current copy at WestHost. I hope this does not interfere with mb.nawcc.org and I cannot see how it could.

--------------- Added 1255727861 at 1255727861 ---------------

How about that!:):D:up:

That one insignificant change seems to have done it. I just sent myself a message using the message board's form mail.

I am very happy. At least until the next shoe drops.

[snakes1100]

Good Job!

Don't forget to fix spamass if the log file still shows its broken in the future if u add incoming email to the server.

[tommythejoat]

I still need to get email integration working, but we have not yet decided if we will really use it with vBulletin.

I am now moving to the next job of setting up the Single Sign On using the .net authentication with some modules provided by our back office vendor.

It turns out the mail problem was not quite as bad as I thought. Some other users who have more lenient mail servers were getting mails just fine all along. It was only us paranoid emailers that have reverse DNS requirements that were being left out.

It looks like over half the new registrations went through OK.

Thanks for the pointers.

[snakes1100]

The reverse record usually points back to the IP holder listed at arin, as long as one is supplied, it don't matter if its reversed back to your domain, i dont recall and email host denying for it not pointing to the correct domain, just as long as it exists for the IP itself.

Yahoo & AOL are the worse.