Apache and Misc problems

I wound up purging, then re-installing apache2 for the 2nd time via the command prompt in ssh. It rebuilt the file 000-default under sites enabled folder in apache2. Doing a regular Webmin module remove for apache, then reinstalling , wont rebuild that file, or it causes a syntax error inside it I think....at least thats what it says when you try to restart apache from Webmin. When you reinstall that apache module, it actualy does not remove anything under the sites enabled folder. I think it may be a good idea for a newer version of the module to ask weather or not to keep them, or remove the sites currently enabled.

I had another problem because my ftp session didnt upload all the files needed for one of the var/www domains...and that took a while before I could figure it out, but with the new 000-default file in there, that is basicly the var/www main index folder, and I cant see a way to get the other domains I have in that folder to load without having that initial virtual 000-default in there. It may be because right now, I am not using a dns to resolve all my domains...I am merely testing via http://serveripaddress/whateverdomain , to make sure all the files are good, and that the mysql system is able to connect - etc. Last night I had the sql working, but seems now that I have removed apache2, it took out the installation of phpmyadmin along with it, so will need to re-install that too.

Definately I will need help with the postfix / sendmail / fetchmail stuff........

Ok, looks like I got the apache2 stuff under control. What I dont understand now is the mail. I have setup mail before with other panels like cpanel / plesk, but something seems to have glitched. I have previous past experience not so great with Qmail, so I elected postfix. Seems we also need fetchmail, so I included that module. Do we need sendmail too? I would think no because if I check the ports I get:

lsof -i :110 lsof -i :25 COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME master 29583 root 12u IPv4 172075 0t0 TCP *:smtp (LISTEN)

So 25 smtp is listening, however seems my pop / 110 port is dead. At one point I had the system accepting the mail but fetchmail would not deliver. Do I need to set up the mail.domain.com over in the bind dns server under each domain I put there? I do have one domain actually set up with a resolving DNS to the site itself, and apache will load it fine, and its mail.domain.com addy resolves to the IP of the server, but when it goes to connect, is says there is no socket, which I guess it because there is nothing listening on port 110. Can anyone help? I did notice at one point sendmail would no longer start, and supplied no error messages, so I removed it as a module....at one point I tried Exim, but its no longer showing up anywhere as a remove or an install possibility. I lije the fact you can switch between all these mail clients. Back in plesk you are stuck with Qmail, and its terrible.

Thank you for that information. Ubuntu 10.4.x

This is the 2nd install of Virtualmin - the first time was the hosting company that installed it, not sure how they did it.When it broke down, I googled for reinstalling , followed the instructions from the command prompt, and scrolled back to my commands to reinstall apache2 from when I encountered problems on the first Vm instaled system. I do not ever remember dovecot being installed at any point by default. Is it a dependent module for postfix?

Here are some of the commands I entered when installing webmin for the2nd time:

.03-1_all.deb

(which I believe failed)

l/libmd5-perl_2.03-1_all.deb

(I think I needed that to do the install)

The rest of these seemed to work :

sudo dpkg -i webmin_1.510_all.deb

(not sure why I had to do this, but I do remember it wasput it in a cdworks folder, probably my mistake)

/webmin_1.510-2_all.deb?use_mirror=cdnetworks-us-1

Then I ran:

sudo dpkg -i webmin_1.510-2_all.deb?use_mirror=cdnetworks-us-1

...which executed from that folder, seemed to install without any problems.

I do recall apache in there as the default, however there was a setup page for the mail where I had to select which system from a drop down box I wanted to use. Options were postfix, exim and I think qmail and some others. At one point it was letting me set exim when it was not installed, but I do remember installing it. Perhaps there ought to be some troubleshooting the team could do regarding that interface someday, because it does not seem you can dynamicly swap from different mail systems if in your configuring, the one you are working with does not work for some reason - seems there was a problem that propogated itself into making no mail client work, save for maybe qmail, which I dont want to try because I have had problems with it before.

Also just tried installing Dovecot , failed !

Installing package(s) with command apt-get -y --force-yes -f install dovecot-common dovecot-imapd dovecot-pop3d ..

Reading package lists... Building dependency tree... Reading state information... The following packages were automatically installed and are no longer required: m4 procmail sendmail-cf sendmail-base libdb4.7 Use 'apt-get autoremove' to remove them. Suggested packages: ntp The following NEW packages will be installed: dovecot-common dovecot-imapd dovecot-pop3d 0 upgraded, 3 newly installed, 0 to remove and 0 not upgraded. Need to get 7805kB of archives. After this operation, 15.7MB of additional disk space will be used. Get:1 http://us.archive.ubuntu.com/ubuntu/ lucid-updates/main dovecot-common 1:1.2.9-1ubuntu6.5 [5510kB] Get:2 http://us.archive.ubuntu.com/ubuntu/ lucid-updates/main dovecot-imapd 1:1.2.9-1ubuntu6.5 [1202kB] Get:3 http://us.archive.ubuntu.com/ubuntu/ lucid-updates/main dovecot-pop3d 1:1.2.9-1ubuntu6.5 [1093kB] Fetched 7805kB in 3s (2582kB/s) Selecting previously deselected package dovecot-common. (Reading database ... 71051 files and directories currently installed.) Unpacking dovecot-common (from .../dovecot-common_1%3a1.2.9-1ubuntu6.5_amd64.deb) ... Selecting previously deselected package dovecot-imapd. Unpacking dovecot-imapd (from .../dovecot-imapd_1%3a1.2.9-1ubuntu6.5_amd64.deb) ... Selecting previously deselected package dovecot-pop3d. Unpacking dovecot-pop3d (from .../dovecot-pop3d_1%3a1.2.9-1ubuntu6.5_amd64.deb) ... Processing triggers for ufw ... Processing triggers for ureadahead ... Processing triggers for man-db ... Setting up dovecot-common (1:1.2.9-1ubuntu6.5) ...

Creating config file /etc/dovecot/dovecot.conf with new version

Creating config file /etc/dovecot/dovecot-ldap.conf with new version

Creating config file /etc/dovecot/dovecot-sql.conf with new version adduser: Warning: The home directory `/usr/lib/dovecot' does not belong to the user you are currently creating. Creating generic self-signed certificate: /etc/ssl/certs/dovecot.pem (replace with hand-crafted or authorized one if needed). hostname: Name or service not known dpkg: error processing dovecot-common (--configure): subprocess installed post-installation script returned error exit status 1 dpkg: dependency problems prevent configuration of dovecot-imapd: dovecot-imapd depends on dovecot-common (= 1:1.2.9-1ubuntu6.5); however: Package dovecot-common is not configured yet. dpkg: error processing dovecot-imapd (--configure): dependency problems - leaving unconfigured dpkg: dependency problems prevent configuration of dovecot-pop3d: dovecot-pop3d depends on dovecot-common (= 1:1.2.9-1ubuntu6.5); however: Package dovecot-common is not configured yet. dpkg: error processing dovecot-pop3d (--configure): dependency problems - leaving unconfigured No apport report written because the error message indicates its a followup error from a previous failure. No apport report written because the error message indicates its a followup error from a previous failure. Errors were encountered while processing: dovecot-common dovecot-imapd dovecot-pop3d E: Sub-process /usr/bin/dpkg returned an error code (1) .. install failed!

But I can see it in the servers area, however it wont start: (Error: ssl_cert_file: Can't use /etc/ssl/certs/dovecot.pem: No such file or directory Fatal: Invalid configuration in /etc/dovecot/dovecot.conf)

Ok, I fixed the hostname -f problem, somehow it had kept part of the old hostname alongside of the new one....and never prompted me to change the option -f part.

Rebooted, and now its listening on the port fine. However any mail sent to the test address gives a relaying error. Also when checking mail, its saying text logins are disallowed. Searched google for that, and saw where in the dovecot conf, that is the default setting. I found 2 config files with that setting = yes, and I changed to = no, but its still giving that message during the login. Rebooted again just to check, and same thing....? Also of note, during the server test for the fetchmail, its saying the mailserver does not equal the address [ fetchmail: Server CommonName mismatch]. I had it as same as server name, and then the actual IP addy of the server and same thing. The domain itself resolves to the server in dns checks....but the name I am using for the hostname is my old servername which has a domain that is still hosted on my old server I am moving from....is that a problem?

Ok, I added the disable_plaintext_auth = no to that file you pointed out, still same problem. That option also was in 2 other config files where I used the file manager search to locate before I did this, and thats where I had also changed that, but still, after doing this, and restarting Dovecot, same thing.

I noticed this file used the brakets { } system, which means there is actual code executing there. I merely placed that line as you pasted, not sure if it has to bein brackets? Because in my fumbling about last night, I came accross a post where someone said you can allow your specific ip address where you log in from such as xx.xx.0.0/16 and the disable_plaintext_auth = no would be valid for just that address range. Could I try that here in the file you pointed out? I cant find the article I was looking at last night anymore in my history.....

Oh, and that test was the fetchmail ' test server ' button, where you can test the server you enter for the fetchmail to execute.

Ok seems disable_plaintext_auth is in lots of files according to the built in search with file manager. I thought there was maybe 3 but theres more. Still getting that same error.

When I first set up the mail, there was kind of like a universal setup page where I could select the kind of system - IE: postfix, Exim , qmail - etc. There was an option for auto detect too. But I cant seem to find this page anymore in the webmin menu? Does the system really work when you switch around from different mail systems, removing and reinstalling them? So far to me seems like there are some problems with that.

Hi Eric:

           Yes, and thanks for all your help. I was thinking of the reinstall last night. 

However, I have got the websites all uploaded and working with apache, and also I installed phpmyadmin, and got it to work with the 2 sites that need sql to operate. It was not easy and one site has quite a large db that meant I had to change the max upload file size to get it to upload. Would I be able to save these settings and sql data locally on the server and still do theinstall.sh you describe ? It seems a waste to have worked on some things, got them all going that are suppose to be more difficult than just the email. I still think this entire problem ought to be investigated by the team for a possible fix in newer versions. It was really easy to break the email installation or misconfigure it. All other panels I have used, never had such a problem like this.

Ok, I guess theres not much else to do. My hosting provider did the first install, so I will ask them if they used the install.sh script the first time. I will have them do a nuke of everything, install Ubuntu fresh , then post here for some more suport.

Ok all set. Fresh install of Ubuntu. Where do I go from here? Will we install as the user or as root? I believe the only root login I can do is the su root.....

Also, is it safe to do any OS updates now before we start with the webmin install, or do we wait until webimin is installed, then do updates?

Hi:

Ok, so we begin with virtualmin? The first time, the hosting company had installed webmin, and from within that I could install virtualmin.....now kinda confused. That link you supplied seems to say that Ubuntu will have to use the manual install.....? The support guy on the server did tell me that they used their own install script for webmin.

I dont have any licenses, so I figure: Virtualmin GPL Downloads

[The Virtualmin GPL installation script is available for Linux and FreeBSD or for Solaris and OpenSolaris 10 and 11.]

No mention of Ubuntu?

The rest of the message under that seems to refer to the manual installation....as a webmin module..

Hi Eric:

Wow, you were right, that went smoothly ! I am doing the post setup now and its asking me :

[For DNS zones created by Virtualmin to be resolvable, the primary nameserver record for each zone must be set to something that can itself be resolved by other systems on the Internet.]

[Primary nameserver} () Skip check for resolvability

Secondary nameservers (optional)

Its got the server hostname they generated during I guess the Ubuntu install in there as the primary nameserver. its a (3) dot address. Since my old server used actually part of a domain I own and hosted on the server as that actual site, I used it as a hostname also with a 3 dot name...and while it worked, plesk panel thought it was a subdomain. Ought I stay away from doing that here in Webmin? The main reason I did it that way was because a Forum 2 email board in php would get alot of bounced emails back from ISP's that would do a reverse match and say that the generic machine name does not have the actual domain. Not all ISP mail servers did this, but a great majority...mostly AOL. So it seemed to improve when I actually put the domain name of that site that had the board actually in the hostname like that. For now, I dont want to deal with that domain just yet, but was wondering your thoughts on it?

Ok, this is the part that is kinda weird. I have 3 domains I will be moving to this new server being setup. The domain that had the mail problem was the one I picked to use as the fqdn last time...it was server1.domain.tld , so if I use that domain now for this server, I cant really get it to resolve without pointing that systems dns to the new server, and right now, I just want to setup mail and http to work with the 3rd dom,ain that is the smallest site, just to get the feel of the system etc. Can that be done?

Other question I have with the mail setup is, do I need to declare a mailserver in the bind dns server for each domain? The hosting company has their own DNS interface outside of virtualmin that seems to take priority. If so, where do I setup the mail server for a domain? Is fetchmail needed with dovecot? Can you supply an example for setting up a mail account for one address? I already setup a user in users and groups as the mail address I want to use, Not as user@domain3.tld, simply as the user name. I made its main group [mail] , not sure if thats needed?

Hi Eric:

Thanks again for the help. I got the virtual domain created, but I learned when creating it in virtualmin, the apache module over in webmin does not always work accurately with the item created in virtualmin. If I change the home directory in the webmin entry, the virtualmin entry wont pick it up. In fact, you cant point it to a different directory if it already exists as a folder, which is frustrating. So I had the domain already working in var/www/domain1 folder with all its files in domain1 folder...then when I setup the same domain in virtualmin , it let me create it, though it already existed...so I had 2 entries. I deleted the one in webmin, tried working with just the virtualmin interface, and I got my one email account going, but now the files for the site are in the webmin directory. Ok, simple move right? Nope. Cant move all those files with the java file manager, PLUS - I dont have permission to even look in the new folder virtualmin created for the new domain. Its made that folder owned by the new name of the domain I guess when I created it, and it wont let you navigate there unless I guess if you login via FTP as that new user. Kind of frustrating how all these permissions are set like that.....

Ok, I thought I understood the new approach as you outlined, but maybe not, or I am just trying to do things that maybe I need to understand better. While I have that domain in the proper location /home/domain/public_html , and the email works, I went to do the same thing for one of my other domains, and it setup in the same format, but if I go to load the first domain, it fails. Its getting information from the domain folder I just entered...and I didnt change anything. This new domain still resolves to my old server, but the one I just got working with the http and email resolves to this server with virtualmin. Is that the problem? One would think that for migration purposes, we can just set up each domain one at a time via http://serverip/the domain , to make sure al the files and stuff and sql work before we actualy do the dns propogation for it all......

Ok, it was because I created the 1st virtual server with just the domain name less the .tld extension. It let me enter without a .tld and I think that was causing anything else to default back to that virtual server. So I changed it to its corresponding .tld, and same for the new virtual server, and so far the 1st domain is coming up ok. To check the second domain, I guess I have to point it to the correct public_html folder....not sure if there is a way to do that, IE: http://serverip/??domain??

Wil read up on that link you supplied, thanks again !

Eric: Hope you are still there for help. I want to have one open directory under /domain1/open and I created the directory then gave it R-W permissions in File manager, but if I go there, its not listing directory contents. I think by default maybe apache does this? If I wanted to permit only this directory from being open could use .htaccess, but I also think apache disables all .htaccess by default, and we would need to change that setting someplace?

Hi Eric:

Ok, tried creating the directory as the admin user for the domain in a ftp session. It worked, but the new folder canot be accessed. its pretty much under public_html/folder for the given domain, and it appears htaccess file does not permit it to be an open directory to the world.On my old server I had this problem with plesk, and there is a setting I think in apache where by default, its disabled the htaccess files, and you can re-enable them, but I do not recall more about it at this time.

On another subject , phpmyadmin.....I notice it can only be installed for each virtual server. Does this mean each virtual server has to have its own database area? I noticed during the troubles before, I was able to instal phpmyadmin via the system command prompt, and have only one main phpmyadmin login, and each domain was able to access the databases after setting permissions and user names etc. So in regard to phpmyadmin, do you recommend using the seperate logins as per the virtualmin setup?