Virtualmin GPL installation on Ubuntu has problems with Postfix smtp chroot?

Virtualmin
1

I’ve spent some time with my Postfix config on Virtualmin GPL, which is installed on Ubuntu-1004-lucid-32-minimal with the install.sh script.

My mail to external domains was not being delivered (mails getting stuck in the queue).
The log showed this error when I requeued the mails:

postfix/smtp[16121]: fatal: unknown service: smtp/tcp
I also had problems logging in from external clients, with the log pointing to issues with sasl:
postfix/smtpd[3361]: warning: xsasl_cyrus_server_get_mechanism_list: no applicable SASL mechanisms
postfix/smtpd[3361]: fatal: no SASL authentication mechanisms

These issues seemed to be a problem with the fact that the default Ubuntu Postfix installation has a chrooted smtp service (see https://help.ubuntu.com/10.04/serverguide/C/postfix.html and http://www.seaglass.com/postfix/faq.html#eruksvc).

After rummaging around with various settings (copying /etc/postfix/services to /var/spool/postfix/, trying to copy sasl files) I now have resorted to getting Postfix working by not running the smtp service chrooted. This seemed to be the best option given my limited Linux and chroot experience (e.g. when I googled, I found conflicting information on whether I should copy or link files, etc.).
I am not aware of having made any changes before I noticed the Postfix setup, so I’m pretty sure the whole installation was default when the problem arose.

Is this an issue that could be adressed in the install.sh script? Or is it the Ubuntu packages that are not getting set up correctly on installation? Looking at the Ubuntu Postfix description, sasl should have worked out of the box, and one would expect the same for postfix/smtp?
As far as I can test this on a local virtual machine (It is based on a Ubuntu 10.04 server, though), all these issues are not present there. I’m trying to find the root cause, so I can address it properly instead of just removing the chroot.

2

Mmm, that’s an unusual problem. Did you notice any errors during the installation? There’s normally no problems in using Postfix after an installation.

You aren’t by chance on a VPS, are you? If so, how much RAM do you have? And do you know what type of VPS it is (OpenVZ, Xen, etc).

-Eric

3

Indeed the server is a VPS (KVM, afaik; /var/log/kern.log contains kernel: [ 0.000000] Booting paravirtualized kernel on KVM)

As for memory:

free -m
total used free shared buffers cached
Mem: 497 437 59 0 7 218
-/+ buffers/cache: 211 285
Swap: 2053 42 2011

That seems reasonable to me?
4

Hmm, yeah, your RAM and such all looks good.

I guess I’m just a little puzzled with why that hadn’t come up earlier. There’s a lot of folks running with Ubuntu 10.04 systems. There a handful of older posts with that problem, but I didn’t see any recent ones.

I’ll fire up a new Ubuntu 10.04 installation and do some testing to see what happens.

Thanks for the heads up!

As far as how to fix it – this fellow here ran into a similar problem with Postfix in a chroot a couple of years ago, and detailed what he did to solve it (about halfway through the thread):

http://www.virtualmin.com/node/11479

-Eric

5

Something has happened that fixed this issue for me, but I do not know what, unfortunately.
The log no longer shows these errors, the smtp; smtps and submission Postfix services are now all running as chrooted services .

My server did reboot due to hardware issues that my VPS provider had, so maybe it was just that one of the services needed a restart (I never manually restarted the saslauthd before, only postfix)

Thank you for the help so far!
I’d say that you can ignore my problem instead of spending time investigating: it is more likely to be my own fault than something in Virtualmin or the install script.