Anyone interested to join forces to create our own hosting control panel?

[ConceptLane]

I currently have something in place, but the interface sucks (I'm no web designer at all).
If there are any PHP programmers, web designers, security experts & Linux experts around, and is open to the idea of joining forces to create a panel for own use (or even release as open source), drop a note here!

I started off writing that for a Win2K host with xampp as the main service, and then I dropped Win2K to move on to Ubuntu, and ported things over.

Some of the pointers I have in mind:
1. No Linux user accounts, purely virtual users
2. For simplicity, single server multiple services, or if ambitious, we can even develop for multiple servers multiple services
3. Single install script to install all necessary services and configuration
4. Primarily Debian/RH/FC/CentOS compatible
5. Configure:
-- Apache2
-- MySQL
-- Bind9
-- XMail / Postfix + Spamassassin + IMAP (optional?)
-- PureFTPd
-- Web Filemanager
-- Awstats
6. MySQL-based accounts
7. Automated/configurable billing system
8. Bandwidth controls using mod_cband (either throttling &/ limiting)
9. Support Tickets
10. Reseller accounts
11. Promotions and offers
12. Payment gateway support (optional)
13. cron scripts to do whatever necessary work
-- backup
-- clearing of session files
-- updating of awstats
-- checking of user space usage (no hard limits due to no actual accounts) and warning emails
14. Direct install scripts?

Some of my friends have laughed at me for trying to re-invent the wheel, but what's the hurdle? I'm sure everyone has their own ideas of how to improve the hosting panel they currently use. Otherwise the other control panels wouldn't keep surviving. Why should only US/EU based control panels be the ones widely used?

I'm thinking a team of 5 - 6 would be good, with 2 programmers 1 designer 1 services expert and 1 security expert (both system and PHP). True-blue MadeInSingapore (for what it's worth).

If there's a market for it, good, can even turn it into a marketable product. Or be hip and cool and release it under GPL. ;p

Alternatively, a control panel solution for one of the many virtualisation technologies, since that looks like the way forward.
Xen/OpenVZ look like stable & growing technologies for the next 3 - 5 years.

[j_lynx]

Hi. I'm opened for discussions for this matter. Please feel free to PM me.

[ConceptLane]

Good to see someone else interested... but you mentioned you're a developer of DTC?

[geekonetwothree]

I will be quite interested, what skillset do you need?

[ConceptLane]

As stated in the first post actually...
An ideal scenario would be to come up with a package that can configure a clean Ubuntu installation with the necessary packages and thereafter make whatever modifications are necessary (without having to rely on HOWTOs and what not) for the control panel to be ready for configuration via the web interface.
A lot to put on the table if we do get this going.

[geekonetwothree]

Base on Ubuntu or Debian would be very nice.

Originally Posted by YorYor

As stated in the first post actually...
An ideal scenario would be to come up with a package that can configure a clean Ubuntu installation with the necessary packages and thereafter make whatever modifications are necessary (without having to rely on HOWTOs and what not) for the control panel to be ready for configuration via the web interface.
A lot to put on the table if we do get this going.

[alanwoo]

Hi,

I have a idea from a difference point of view. The objective to have a control panel is to simplify the end-user experience on using the hosting services. eg. create email account, change password, web-based file manager, configuration settings etc.

I find that existing control panel is difficult for end-user to use it.

Why not develop a front-end application, with the objective to simplify the end-user experience which will be install on end-user PC, that allow the end-user to configure their website/email setting from this front-end application(control panel the same way), and synchronize these setting with the server.

The backend technology can be based on web-services, xml, soap etc.

This front-end application can compatible to works with major control panel (eg. cpanel, directadmin, plesk etc).

Alan

[geekonetwothree]

this is a brilliant idea! Why not we do something like that?

Originally Posted by alanwoo

Hi,

I have a idea from a difference point of view. The objective to have a control panel is to simplify the end-user experience on using the hosting services. eg. create email account, change password, web-based file manager, configuration settings etc.

I find that existing control panel is difficult for end-user to use it.

Why not develop a front-end application, with the objective to simplify the end-user experience which will be install on end-user PC, that allow the end-user to configure their website/email setting from this front-end application(control panel the same way), and synchronize these setting with the server.

The backend technology can be based on web-services, xml, soap etc.

This front-end application can compatible to works with major control panel (eg. cpanel, directadmin, plesk etc).

Alan

[ConceptLane]

That's an interesting idea too! It's definitely something which isn't addressed by the current available solutions.

However, the caveats I can see are:
- not sure users will be keen to install software on their PC
- even more troubleshooting in future should something go wrong while deployed (client end software errors?)
- having it compatible with the various panels would be a very big challenge
- even more developers needed, especially Win32/GTK programmers.

[Lance]

I am interested but I am currently based in australia. So let me know via pm. I normally check it once a week

[geekonetwothree]

So are we going ahead?

[alanwoo]

The reason to use a application installed on user pc compare to web-based/ web-browser is because of the functionality available via web-browser is limited.

Although currently we have "ajax" technology to resolve some limitation on web-browser, it is still quite not user friendly.

If you have time, try check out the new "Windows Live" application, it is a family of "Windows Live Mail", "Windows Live Messenger", "Windows Live Gallery", "Windows Live Writer". These application installed into your PC, allow you to synchronize your email, blog or gallery that reside on Live.com server.

The direction is clear that, after years of trying to be fully "web-based"(thin client) solutions, they decided to develop a thick client application to compliment the web services.

This is same as what is "mac.com" trying to achieve. Check out mac.com if you have time too. Mac.com allow you to synchronize you address book, calendar, email, gallery, webpages from the mac application installed in your mac pc/notebook.

[ConceptLane]

Agree that it may be coming to the change in the cycle, but I believe that web clients will remain.
Personally, I would rather we design and build a CP that perhaps caters for XML-coded updating, and let other budding software writers do the executables+GUI.
I believe if we publish such a CP with an updating API and declaration of security measures, the interest will be there.
If we get a real Win32/GTK programmer onboard, we can even offer the software as part of the bundle for webhosts to place their logos and contact information etc.

geek and Lance, if you guys are keen, sure! Mind sharing your skillsets? We can arrange some form of MSN/IRC netmeeting to further align our ideas and perhaps do some splitting of work.

[ConceptLane]

Anyone else keen to participate?

[Darylngch]

Originally Posted by YorYor

Anyone else keen to participate?

Well... If you need any beta tester, i can help you to. In terms of programming it, I find myself too rusty to help you (Been a long time since i last touch on programming).

[ConceptLane]

After some inertia, prelim work has begun.
Will take a long time to catch up completely to the features present in cPanel, but for basic config of domains, subdomains, mysql dbs, email accounts etc should be more than capable. I'll need to take time to revamp code as well, so it will take some time. Anyone interested can keep checking back here for updates. I'll probably set up a site for it to open up the source once something is ready for alpha testing.

First, infrastructure.
Dev platform: Ubuntu 8.04.1 32-bit (testing in jeos environment)
Packages:
- Bind9 for DNS
- Apache2 for client vhosts, with php5 and mono modules, and non-official cband module for bandwidth restriction
- MySQL5
- PHP5 (apache+cli+cgi versions) with curl, ming, mcrypt, mysql, gd, sqlite3 extensions
- Lighttpd for control panel specific access
- For email, I'm not going to be using the typical postfix+dovecot or spamassassin, as I've currently deployed xmail+assp to fairly good results. So it will be xmail with manual upgrade to 1.25 and assp 1.3.3.10 as the anti-spam proxy
- Awstats with free geoip library for site statistics
- Pureftpd for FTP
- Clamav for virus-scanning by ASSP
- phpmyadmin
- telaen for webmail

Design principles:
1. 100% virtual accounts
2. All services will run as their own default user, but will be added to a panel-specific group for write access
3. All data files to be moved into a directory in /home/<panel>, as well files belonging to vhosts
4. PHP to employ open_base_dir directive to limit access
5. Bandwidth restriction is by speed/throughout rather than amount of data transferred, with an optional to disable and limit by amount of data transferred instead. Plan to impose limits to slow it down to a trickle instead of hard-blocking sites once data transferred is exceeded.

Identified Caveats:
1. Lack of hard-limiting quota system - pureftpd can limit when accessing via ftp protocol, but file uploads etc may cause over-usage.
--> planned action - implement of soft limits, pureftpd .quota limits to be calculated at specific time daily, and emails to be sent to both admins and user to warn of over-quota - "a more human and tolerant approach" but requires extra diskspace available and creates a bit of workload for admins, but users will not be troubled by hard limits

Still unknowns:
1. "jail" perl without running apache in a jail itself
2. no experience with mono and necessary config to ensure security

Probably more points to raise, but I'll add them on as I go along.
Feel free to reply if anyone has any suggestions or pointers.

[alanwoo]

Hi,

Nice to see there is some real works already started, as I am currently doing some migration of web host account from one server to another due to upgrading of OS(operating system), I would like to contribute some experiece in this area that may benefit for this developement.

We have been migrating hosting account from servers to servers since we started from RedHat9->Fedora1->Fedora4->CentOS4->CentOS5, or Windows 2000 Server->Windows 2003 Server->Windows 2008 Server. And migrate application from php4->php5, mysql3.23->mysql4.1->mysql5

This process is endless.

We have managed to solve the hardware migration issue by using Virtualization technology, so migrating virtual machine from difference server hardware will still works. But this is not so lucky when migrating between difference OSes or application version.

With the new 64bit OS/servers, IPv6, some application just wont be able to move ahead.

There probably need an additional layer to address this kind of issue. Java is one type of the solutions to address the application portability issue between OSes.

For hosting solutions, we probably need something similar to Java VM, that create a new layer to isolate hosting services with the underlaying OSes or applications. Hosting services should be able to continue to function even the underlaying OS or application have upgrade to another newer version.

End-user always face this dilemma when hosting company upgrade their OS/software and breaks their existing codes, and end-user will have to forks out additional money to get web developer to fixed the code for them.

Alan

[ConceptLane]

The most "painless" migration in terms of 100% hardware abstraction is probably VMware/Virtualbox. But even then, the OS is one that you probably can't break away from, except from your suggestion of Java. But I'm not so keen, because the overhead of the Java layer places additional load. I was never a big fan of java, ever since it was part of my curriculum long ago.

But I think no matter what, it'll be a never-ending cycle.
Main thing is to get it set up on a long term support OS like the Ubuntu LTS series, and then try to place it on a moderately capable server to last the same duration (3-5 years).

I even thought about having it capable of controlling multiple servers, but it's a bit of a headache just thinking about it. Like the installation script prompting to install AIO server, or individual DNS/Email/FTP/MySQL/Web1/Web2 servers etc. But that would require a fair amount of expertise and hands-on control from the installer, must do manual setup of child/slave servers and IP addresses etc... argh. Would such a feature be in demand? Any CP out there that does this already, allowing separate IP addresses pointing to different server types, with one single control? I think this is something I'd like to be able to do...

[ConceptLane]

As mod_cband is deprecated, bandwidth limiting will be switched over to use mod_bw instead.

Also, would integration with tomcat be a necessary thing to have? Currently near 80% to getting the installation script completed. So better to consider whether to integrate tomcat as well if the usage/demand is high.