Reinstallation

You should never run the install.sh script on a system that is already in service. It is for installation. If there were errors during installation, sometimes it is useful to run the uninstall process and run install.sh again after correcting whatever problem caused the error(s). But, if you install weeks ago, and have been using it...it is not a good idea to run install.sh again.

The virtualmin-base package does include a step that attempts to open all of the necessary ports on the local firewall, but if it failed once, it'll almost certainly fail again. It cannot accommodate custom firewalls very well (though it handles most of the common ones--the Red Hat default firewall, for example, can exist or not, and the install script works either way...but oddball stuff like APF are broken by design, and don't interact with other firewall rules politely).

So...the question is: Do you have any Virtualmin accounts you want to keep? If not, you could run:

install.sh --uninstall

And then run install.sh again. Watch carefully for errors, and bring them up here if you see any, and we'll try to help you deal with them.

If you haven't changed any configuration in Virtualmin that you want to keep, you could also just re-install virtualmin-base. This won't (or shouldn't...make backups anyway) break any of your Virtualmin accounts, but will restore the default configuration of many services and try to enable quotas and such.

To do that:

rpm -e virtualmin-base
yum install virtualmin-base

Or:

dpkg -r virtualmin-base
apt-get install virtualmin-base

This is still potentially disruptive, so you don't want to do it on a production box.

<div class='quote'>Did it affect the rest part of the script?</div>

Yes. Looks like nothing in virtualmin-base actually ran, if that's the entirety of the log between &quot;install virtualmin-base&quot; and &quot;Install of virtualmin-base completed&quot;. There should have been a dozen or so &quot;Configuring...&quot; lines in there. Oh, wait, you've sent the SSH log. Look in /root/virtualmin-install.log, as well...I believe we're logging everything that really ought to happen.

I'm looking into this particular issue...sounds like maybe the default aliases file on CentOS now contains something that it didn't when we built the virtualmin-base package. I'll probably need to accommodate that.

<div class='quote'>After figuring out problems of Mail, Web, DNS for a few days, I now can create a web site and upload files to it. But I still can't connect to mail server at port smtps(SSL smtp) due to missing SSL cert.</div>

virtualmin-base doesn't configure smtps. Because it requires a certificate, and many users don't have one, or have it in place on the system, this is a step that is saved for later. I've answered your thread about this with more detail about how to setup a cert for Postfix.

<div class='quote'>I also found that were some misconfiguration in the default Postfix config like mailbox_size_limit and message_size_limit.</div>

Hmm...I was thinking that only adversely effected the Debian virtualmin-base. I'll have to roll a new one for CentOS/RHEL, too. As mentioned in the thread about this particular problem, just remove those two options from /etc/postfix/main.cf, and this error will go away. Will be fixed in the next version.