Proxmox 1.4b - OpenVZ containers - emails not being sent

hisaltesse · Oct 8, 2009

I migrated a few OpenVZ Centos 5.2 VPS from a Proxmox 1.1 server to a new Proxmox 1.4b server (Lenny).

Some of the VPS were running an application that sends emails to users via an SMTP server configured in the app.

Now users are no longer receiving any emails.

I looked at the exim_mainlog of one of the VPSs and here is what I see.

Any idea what is going on?
- the fact that many vps suffer makes me wonder if this is a proxmox issue

2009-10-07 21:06:46 Start queue run: pid=19219
2009-10-07 21:06:46 1MvW97-0008II-8V Message is frozen
2009-10-07 21:06:46 1Mv9fK-0000di-6S Message is frozen
2009-10-07 21:06:46 1Mv0cj-0001rj-QN == root@test.mydomain.com R=localuser T=local_delivery defer (-52): Retry time not yet reached
2009-10-07 21:06:46 1MvN6I-00015A-3N == root@test.mydomain.com R=localuser T=local_delivery defer (-52): Retry time not yet reached
2009-10-07 21:06:46 1Mv0cq-0001tV-R9 Message is frozen
2009-10-07 21:06:46 1MvN6Q-00016x-EE Message is frozen
2009-10-07 21:06:46 1Mufi9-0005EQ-Ql Message is frozen
2009-10-07 21:06:46 End queue run: pid=19219
2009-10-07 21:07:49 no IP address found for host localhost (during SMTP connection from [127.0.0.1])
2009-10-07 21:12:51 no IP address found for host localhost (during SMTP connection from [127.0.0.1])
2009-10-07 21:17:51 no IP address found for host localhost (during SMTP connection from [127.0.0.1])
2009-10-07 21:22:52 no IP address found for host localhost (during SMTP connection from [127.0.0.1])
2009-10-07 21:28:46 no IP address found for host localhost (during SMTP connection from [127.0.0.1])
2009-10-07 21:32:57 no IP address

dietmar · Oct 8, 2009

That looks like an MTA configuration problem.

hisaltesse · Oct 8, 2009

Dietmar,

Could you please be more explicit? What is an MTA configuration problem?

Could you please recommend some actionable steps to do in order to remedy the problem?

dietmar · Oct 8, 2009

hisaltesse said:
Dietmar,

Could you please be more explicit? What is an MTA configuration problem?

MTA is "Mail Transport Agent" - exim in your case. Seems there is something wrong with your exim configuration.

- Dietmar

hisaltesse · Oct 8, 2009

So would you think this is at the proxmox level or at the VPS level?

And how do you explain that it was working before until migrated to the new proxmox server?

And what would you think could be wrong with the exim config? Where should I start looking at?

dietmar · Oct 8, 2009

hisaltesse said:
So would you think this is at the proxmox level or at the VPS level?

or maybe a combination.

hisaltesse said:
And how do you explain that it was working before until migrated to the new proxmox server?

maybe you have done a stop/start (which rewrites some configuration files)?

thefool808 · Oct 8, 2009

hisaltesse said:
So would you think this is at the proxmox level or at the VPS level?

And how do you explain that it was working before until migrated to the new proxmox server?

And what would you think could be wrong with the exim config? Where should I start looking at?

What does the VM's /etc/hosts look like?

hisaltesse · Oct 8, 2009

I have indeed done a stop/start... I was not aware of the differences between that and shutdown.

Please explain that difference; how does stop/start changes the configuration? Which config does it change and how can I remedy it?

Does this apply as well when you do a "non-live" container migration? (I migrated some containers without checking the live migration check box because of course the live migration does not work...

Let me know.

Thanks.

dietmar · Oct 8, 2009

hisaltesse said:
I have indeed done a stop/start... I was not aware of the differences between that and shutdown.

nobody claimed that.

hisaltesse said:
Please explain that difference; how does stop/start changes the configuration? Which config does it change and how can I remedy it?

Each 'start' rewrites a bunch of configuration files (for example /etc/hosts) - please consult the openvz manual for details.

hisaltesse · Oct 8, 2009

thefool808 said:
What does the VM's /etc/hosts look like?

It looks like this:

# Auto-generated hostname. Please do not remove this comment.
1xx.xx.xx.xx test.mydomain.com test test.mydomain.com
r

thefool808 · Oct 8, 2009

Edit it to look like this:

Code:

127.0.0.1 localhost.localdomain localhost
# Auto-generated hostname. Please do not remove this comment.
1xx.xx.xx.xx test.mydomain.com  test test.mydomain.com

Verify that everything works, then restart the VM and test it again.

hisaltesse · Oct 8, 2009

Could you please explain what this is supposed to do? Because all openvz containers seem to be created and running without that line that you suggested.

thefool808 · Oct 8, 2009

In the end it is still Exim specific, as it seems Exim is trying to resolve localhost. Maybe if you wanted to not touch /etc/hosts, then you could change the Exim configuration to resolve to your hostname. I don't know Exim enough to suggest how to do that. A simple fix might be to update your /etc/hosts. Again, this is just a guess on my part.

My debian OpenVZ VMs are definitely created with that line in /etc/hosts.

dietmar · Oct 8, 2009

What centos template did you use to create that VMs?

hisaltesse · Oct 8, 2009

I tried the /etc/hosts change and rebooted but it did not solve the issue.

-- Dietmar
I am running:
CENTOS 5.3
Template centos-5-standard_5.2-1_i386.tar

thefool808 · Oct 8, 2009

I guess the following part of the log might not have been related to Exim not being able to send emails.

Code:

2009-10-07 21:28:46 no IP address found for host localhost (during SMTP connection from [127.0.0.1])

You may want to try and resolve an MX record for the email domains of the emails that are not being sent from within the VPS and then try to ping that server.

There is hopefully more log messages detailing Exim actually trying to connect to another email server and failing to either connect or failing during the SMTP conversation. I see in the log you posted a message like this:

Code:

1Mufi9-0005EQ-Ql Message is frozen

But I want to know, why is the message frozen? Maybe searching the Exim main log for that message ID will do it.

Code:

cat exim_mainlog | grep 1Mufi9-0005EQ-Ql

dietmar · Oct 9, 2009

hisaltesse said:
I am running:
CENTOS 5.3
Template centos-5-standard_5.2-1_i386.tar

I just created a VM with that template, and /etc/hosts includes a line for localhost - seems you modifyed some files manually (or another program inside the VM modified it).

hisaltesse · Oct 9, 2009

I personally did not modify anything manually.

The only thing that was modified was the host /etc/network/interfaces to add more IP ranges.

The VPS I was referring to runs cpanel. Unless cpanel modifies that file?

However I added that line and did not get this issue solved.

hisaltesse · Oct 24, 2009

Alright guys I finally found the issue.

This had nothing to do with Proxmox....

a vps user changed their mail server while we were migrating the vps to the new hardware and that is why the emails were not going out...

Nothing wrong with proxmox with respect to this issue... Thanks for your support.

Proxmox 1.4b - OpenVZ containers - emails not being sent

Well-Known Member

Proxmox Staff Member

Well-Known Member

Proxmox Staff Member

Well-Known Member

Proxmox Staff Member

thefool808

Guest

Well-Known Member

Proxmox Staff Member

Well-Known Member

thefool808

Guest

Well-Known Member

thefool808

Guest

Proxmox Staff Member

Well-Known Member

thefool808

Guest

Proxmox Staff Member

Well-Known Member

Well-Known Member