I installed Virtualmin/Webmin using the 6.0 script on a Centos 7.3 VPS and Apache was working. I then disabled Apache and installed NGINX. I ran into a glitch - the etc/initd/nginx startup file was missing. That required copying a file to the directory and giving is root:root ownership and 0755 permissions. I have run into that problem with the older install script as well.
I have not fully tested the fresh intall and will report if issues crop up.
The new install script installs PHP 7 in addition to PHP 5.4. However, Virtualmin fails to properly recognize PHP7 and uses PHP 5.4. That version is no longer supported by several plugin scripts including PHPMyAdmin, Joomla, and Wordpress. The Vitualmin>PHP Versions shows an error message similar to the earlier install script:
Error
Website options cannot be edited, as no Apache virtual host for cloud4g.com port 80 was found! That problem has been lingering over the past few years.
That means that the remi repo must be installed to upgrade PHP beyond v5.4 or deleted and hope Virtualmin will then recognize PHP7.
I think the baseline server many people want will have PHP7, the option for NGINX, and fewer work-arounds. PHP 7 and NGINX are needed because performance improvement is substantial
I installed the latest remi repo including repo for remi PHP71 and enabled in remi PHP5.6 and in remiPHP71. YUM update installed PHP7.1.8 which replaced PHP5.x which I don’t need.
The dependent scripts PHPMyAdmin and Joomla CMS were installed and work. I have not tested other script installers but expect they will work.
If you want nginx, you should use the --bundle LEMP option when installing. That installs nginx instead of Apache. It’s still new, and quite untested, but if folks will test it and report problems, it’ll get better quickly.
As for /etc/init.d, that’s not a thing on CentOS 7. CentOS 7 uses systemd, there should never be an init.d file for any service (well, we still ship one for Webmin, but that’ll be fixed in the 2.0 Webmin branch coming up soon). But, you shouldn’t copy an initscript for nginx. There is a systemd unit file for systemd, which handles starting and stopping and the like. In my tests, this part of things is fine. But, if you have problems starting and stopping nginx using the usual system tools or Webmin/Virtualmin, let me know.
The Remi PHP packages come in two flavors; one that replaces the standard PHP, and one that exists beside it (those are the ones in SCL). We prefer the ones that do not replace the system PHP packages. If there’s a problem with them, we’ll get them fixed. We just need to know about it. It should not be necessary to replace the system package in order to get Virtualmin to recognize and work with another version (but, because there’s infinite possible places for packages to be installed, we don’t currently support every combination of packages out there…we kinda hope to discourage folks from installing random PHP packages, which is why we’re setting up our preferred packages out-of-the-box when possible, so we at least are working with only a couple dozen possible combinations instead of hundreds of possible combinations).
Anyway, PHP7 and PHP5 are both correctly detected on my two CentOS 7 test systems, so I’m not sure if your experience is common. I’d need to reproduce it to fix it, I think. Anybody else seen PHP7 not detected after a fresh installation of Virtualmin 6 on CentOS 7?
Thanks Joe! This sounds like a great improvement. I did not know about the installation options. That helps a lot.
Found a bug in the new installer on Debian 8 jessie:
Line 676: run_ok “apt-get install apt-transport-https lsb-release ca-certificates” “Installing extra dependencies for Debian 8”
this line will get the installer stuck because the apt-get command requires confirmation, I manually added “-y” after the apt-get command and it worked for me, it’s like: run_ok “apt-get -y install apt-transport-https lsb-release ca-certificates” “Installing extra dependencies for Debian 8”
Yep, that one is fixed in github, but I guess I haven’t rolled it out to the repos since that fix went in. Good catch and good fix.
I reinstalled Centos 7 full image install on a LXC VPS with 2Gb RAM, 40GB disk, did a yum update. Then I installed Virtualmin using the --bundle LEMP option. The installation was reported successful. The only warning (red) was that quotas were not implemented.
Then logged into https:/mysite.tld:10000 and went through the initial setup. That finished, the screen prompted to recheck the configuration. That results in this message:
"Your system has 2 GB of memory, which is at or above the Virtualmin recommended minimum of 256 MB.
BIND DNS server is installed, and the system is configured to use it.
Mail server Postfix is installed and configured.
Postfix is configured to support per-domain outgoing IP addresses.
PHP execution via fcgid requires the Apache mod_fcgid module
… your system is not ready for use by Virtualmin."
I then tried to install the mod_fcigd module but is came back as already installed and the lattest.
Joe, I volunteer the mentioned VPS for a few days to use for trial testing of the script. Let me know and I will provide SSH login information. If you want to use different OS, I can install Fedora 17 32 or 64, Suse 12.2 32/64, Scientific 6 32/64, Debian 6 32/64, Ubuntu 13 32/64,
I’m on Debian 8 and Virtualmin 5.99, (LEMP: Nginx 1.10.3 + PHP-FPM and MariaDB 10.1).
What I have to do to go to VM6?
I have to simply launch virtualmin config-system?
The Virtualmin install script is meant for initial install of the Virtualmin/Webmin stack on a clean/fresh supported OS. Using the script on an existing Virtualmin or another application stack can be expected to cause problems and is not recommended.
This is not an upgrade of an existing Virtualmin/Webmin server. That occurs through regular updates.
Hello Joe,
Have all the reported problems been fixed? I Installed Virtualmin last Sunday in fresh mininal install of Centos 7, and got several problems:
- php files being downloaded and not run
- clamAV does not work
- severalpostfix problems like authentication (plain text didn’t work at first), error on configuration variables not found ($mua_helo_restrictions,
$mua_sender_restrictions), SMTP auth I still can’t make it work, when sending emails I get a wrong password message, and it is correct.
I’ll have to reinstall the server I installed last week, but now I’m afraid to use V6…
Is it safe to use V6 on a production environment?
Thanks
- php files being downloaded and not run
Fixed a couple days ago. (And this only applied to fcgid execution mode…FPM and mod_php should have worked, anyway.)
- clamAV does not work
Um…I don’t know of any reported problems with ClamAV. You’re gonna have to report a problem if there is one, since I don’t know of one. 
Note that the --minimal mode installation does not include the clamd (or spamassasin) server. If you want local mail processing, you have to install the full Virtualmin installation. That’s pretty much all --minimal installation does: Removes the huge clamav server and spamassassin (not as huge as ClamAV but still big) from the default configuration (a few other packages get pushed out to suggests rather than recommends, but you won’t notice those probably).
- severalpostfix problems like authentication (plain text didn’t work at first), error on configuration variables not found ($mua_helo_restrictions,
I fixed authentication. I haven’t seen any reports about mua_helo_restrictions (and we don’t touch that option or require it, so I have not idea what it’s about). But, if you have unique problems you have to report them, or they can’t get fixed.
Virtualmin 6 is a better installation choice than Virtualmin 5. It fixes several issues with the Virtualmin5 installer, and adds a number of additional features by default. I didn’t rebuild the whole installer for fun…working with POSIX shell scripts isn’t fun. I did it because the old one had a lot of problems that we continually papered over with hack after hack to make it mostly work most of the time. New one had some bugs, but, I fix them as they’re reported. And, the new one almost always allows one to fix things without a full re-install (whereas old one didn’t have a configuration management tool, so every fix required a new install or a bunch of manual fixing).
So, report problems you find, when you find them and I’ll fix them and tell you what you need to do to apply the fix(es) to your server. Usually it’s a matter of updating the virtualmin-config package and running a single command. For example, one can fix the broken saslauthd configuration that prevented SMTP authentication by running: virtualmin config-system --include SASL
Give me a few more days to knock out any remainder quirks with the installer and Virtualmin 6. I will provide a script to perform the following steps: Switch repositories from the old ones to the new VM6 ones, install virtualmin config-system command, and optionally add some of the new features (only when explicitly told to do so, as those features might conflict with existing configuration).
For now, the virtual-server module 6.00 is in all repos. So, you can already get that, and it is safe to roll it out to any existing Virtualmin (and should be deployed to any Virtualmin system).
@Joe, I reported Clamd problems in this thread:
https://www.virtualmin.com/comment/781134#comment-781134
I used the VM5 installer, as Mail doesn’t work with the VM6 so far
The latest install script, v6.05 works for the standard Apache, PHP5.4/PHP7.0 installation on an LXC virtual server without modifications! The script also works as experienced on two different LXC VPS provider’s platforms using the --bundle LEMP option that installs NGINX instead of Apache. However, this requires post-install change of Virtualmin>features and plugins. The install leaves (Apache) SSL Website, AWSTATS Reporting, and Protected Directories checked as enabled. Those must be manually de-selected and the features and plugins saved. At that point, Virtualmin>Re-Check Configuration shows no errors, “your system is ready for use by Virtualmin”.
Where is promised separate port settings for NGINX web server? I can’t enable both Apache & Nginx on one virtualserver (in reverse proxy scheme) , because nginx gets port number from Apache settings and shows the error:
Failed to modify server : Nginx cannot be enabled for a domain that also has an Apache website
This was be reported some month ago, and promised that in Virtualmin 6.00 this problem will be fixed. But seems this is not fixed. We need separate settings for NGINX port number!