Ubuntu Home Server

[Matthew]

I'm posting a new topic, forking from a post about the Ubuntu Home Server project that was offtopic in "Re: Media Storage Idea", copying that original post. Please continue discussion of it in this topic.

This project is coming along nicely:
http://www.ubuntuhomeserver.org/

I hope that both of these projects do not try to do to much (focus on getting the things they can perfect) but in the future could work together to make one hell of a system.  I like the security of the UHS, the GPL, and its data storage capabilites.  I like LMCE to play media and automate the house.  I worry about security in LMCE.

That UHS project is designed to make Ubuntu 0804 include more features for the home, including simpler GUIs for administering services for a networked multimedia home. Since LMCE is supposed to be a further packaging on top of Ubuntu, there's no reason they can't work together, if designed properly modularly. They each need to be able to omit their specific chosen components in favor of workalike substitutes that meet the same API. And their installers need to be complementary.

So UHS needs to be able to use the LMCE component where the two projects present alternatives. And LMCE needs to be able to install as packages onto UHS the way it currently does on Kubuntu, but also select (and substitute) between packages where they conflict between the two. The Debian/Ubuntu APT installation system accommodates that requirement well, so if it's used properly we can do it. But the main upgrade to LMCE is just being able to upgrade an LMCE's underlying Ubuntu against the Ubuntu repository systems, including a new Ubuntu version, without trashing the rest of LMCE. LMCE 0710 is supposed to allow that. If it does, integrating the two complementary systems will be fairly straightforward.

And beneficial. Mainly because there would be a larger, unified community of developers, who could work together on the unified set of tasks. And because the "conflicts" in each project's selections for functions also offer more choices, therefore appeal to a larger group of users who might prefer one or another choice. And also because the requirements for the integration would improve the overall robustness of each project, more integrable not only with each other, but also with other possible systems, either current or to come in the future.

I recommend people participate in this discussion in each project's forums. The UHS project started in reference to LMCE back in May, so we're catching up with them in cooperation. Even if the UHS project dies, it's worth looking at for more features and robustness as a home solution. Thanks for bringing it to our attention.

[Matthew]

Well to clarify what I was hinting at with my post.

I have yet to install LMCE as I am sifting through all the hardware requirements and specs.  Going to order and build my system when I am comfortable with my hardware choices.  I need to manage a sat system, Over the Air broadcasts, 4 cable boxes, and want to implement home automation,share media in-home and over the net, have some control from the net of LMCE.

Ran across this post about storage.  I foresee this being a problem for me also.  Currently I have been running a regular ubuntu 7.10 server trying to share my media both in home network and over the internet.  I installed and use Gallery2(http://gallery.menalto.com/)  for pictures and Ampanche (http://www.ampache.org/)  for music (very exicited about KDE4 as AmaroK2 will have an Ampache interface and work on linux and windows).

So as I started researching I found (without much trouble) LMCE and UbuntuHomeServer.  I plan to run a total home solution with LMCE and then have UHS server provide services such as store my media, provide better sercurity to LMCE by providing a proxy and additional firewall, host a website that will allow me to access my media on the road and also to schedule recordings and events to LMCE.

This is all very exciting to me and I am still learning all the ins/outs of LMCE capability.  But I do know that I want my in-home systems removed as far as possible from the internet while still providing the control I want from the internet.

This is a wonderful project and I cant wait to see how KDE4 will bring an even more sexy user experience to the project.

What could be interesting would be a feature comparison grid betwen UHS and LMCE. That would show what features each has in common, which specific package is used by each to deliver the feature, and where they're purely complementary. It would be the roadmap to making them interoperate, and indicate if indeed that is a good idea. And it would be the least technically demanding part of such a project.

[1audio]

Most of the features mentioned are already in LMCE. the biggest security issue seems to be having the hardware at the firewall. An external firewall would address the security issues probably better since no accidental hole could give direct access to the protected content. If you use UHS the same shared hardware issue is still there. UHS has the benefit of doing a lot less so it may be better constructed to manage in that environment.

[bulek]

I also have same desires. Beside LMCE features, I also need family server, with some internal blogging (wordpress), printer server, some groupware app (perhaps Egroupware), gallery2, etc....

IMHO, the proper way would be to go slowly from UHS with adding LMCE functionality... At the start there is a desire for having reliable and easy to manage home server, and then all the features of LMCE added... But since, both systems are up and running, I think it wise decision to merge those two projects in some most convenient way (as I think now, having LMCE based on UHS would be ultimate solution) or at least expand LMCE with those features that are vital to family houses...

Regards,

Bulek.

[tschak909]

they seem to be more talk than action as of yet.

-Thom

[totallymaxed]

they seem to be more talk than action as of yet.

-Thom

Agreed Thom... I think it will be a long time before they anything thats stable & usable

[Matthew]

they seem to be more talk than action as of yet.

-Thom

Agreed Thom... I think it will be a long time before they anything thats stable & usable

Well, they were targeting a release as part of the Ubuntu v0804 (Hardy) release, so we should have a benchmark of their project's performance by then.

However, if they could use LMCE's existing achievements to achieve some of their goals, they might be able to achieve more by then by focusing on just what's left, as well as the other synergies from the LMCE momentum. And conversely, there's a chance that their goal of releasing on Ubuntu's own release schedule could help LMCE get on that schedule, too (or a short, fairly predictable lag from it, like +1 month).

[ddamron]

they seem to be more talk than action as of yet.

-Thom

Agreed Thom... I think it will be a long time before they anything thats stable & usable

Well, they were targeting a release as part of the Ubuntu v0804 (Hardy) release, so we should have a benchmark of their project's performance by then.

However, if they could use LMCE's existing achievements to achieve some of their goals, they might be able to achieve more by then by focusing on just what's left, as well as the other synergies from the LMCE momentum. And conversely, there's a chance that their goal of releasing on Ubuntu's own release schedule could help LMCE get on that schedule, too (or a short, fairly predictable lag from it, like +1 month).

Heh, and all of us contributors become MOTUs? lol

<Reply to this USELESS comment not required>

[Matthew]

Well the one biggest achievement they could have IMHO for LMCE would be to figure out the LVM - RAID stuff with backup automation.

I haven't looked into deficiencies in LMCE RAID/backup. What would UHS do better than LMCE does?

The rest is academic adding gallery2, Ampache, wordpress, etc to the server.

The trick is not so much just installing those apps in the default distro. It's integrating it with the rest of the GUI. Turning documents and their management into a mediatype like video is, introducing "collaboration" across the LMCE featureset, making their gallery an install-time (or upgrade-time) option vs the default now in LMCE, exposing the rest of the LMCE media and metadata as content easily embeddable from within Wordpress... LMCE isn't just a distro, it's an application and a unique, integrated GUI.

[dragon_788]

Hey guys, just wanted to give you a heads up on a thread I started over at the UHS forums to address this EXACT topic. One thing that I agree with what somebody said is I do NOT want my HOME AUTOMATION server touching the internet directly, that to me means that anybody with a bit of time on their hands could exploit my system and really screw up my home, and if I wasn't there I wouldn't have a clue til I couldn't access the system, I've also heard from well moneyed customers of a home solutions company I work with that do NOT want certain portions of their entertainment systems to even touch the lights/HVAC and other parts that LinuxMCE tries to tie so tightly together.

Anyways, back to the task at hand, finding the "best" solution for storage as well as network management to complement LinuxMCE. Feel free to weigh in on the thread over there or over here, I'll try to keep up with both, one thing I don't want to see is LinuxMCE try to do EVERYTHING, and be good at nothing, I've heard from a few people already that don't like the interface and would really like to see that improved before they even think about selling it. So I'm imagining a trifecta of servers, UHS on the outside/edge, unRAID sitting behind it accessible to both UHS and LinuxMCE and LinuxMCE in the "middle" away from the internet but still reachable. http://ubuntuhomeserver.org/phpbb3/viewtopic.php?f=1&t=381

[tschak909]

and you honestly expect an average home to have _THREE_ servers in the house?

methinks you are a tad out of touch with reality and are living in geekland.

-Thom

[Matthew]

So I'm imagining a trifecta of servers, UHS on the outside/edge, unRAID sitting behind it accessible to both UHS and LinuxMCE and LinuxMCE in the "middle" away from the internet but still reachable. http://ubuntuhomeserver.org/phpbb3/viewtopic.php?f=1&t=381

You can already use an unRAID server with LMCE instead of using its builtin RAID for content (or anything else other than the LMCE OS/apps). Just run Samba on it and expose it as NAS.

One thing that I agree with what somebody said is I do NOT want my HOME AUTOMATION server touching the internet directly, that to me means that anybody with a bit of time on their hands could exploit my system and really screw up my home, and if I wasn't there I wouldn't have a clue til I couldn't access the system, I've also heard from well moneyed customers of a home solutions company I work with that do NOT want certain portions of their entertainment systems to even touch the lights/HVAC and other parts that LinuxMCE tries to tie so tightly together.
(...)
I've heard from a few people already that don't like the interface and would really like to see that improved before they even think about selling it.

One major architectural improvements that LMCE could stand would be more factoring of services to different servers, and clustering of them. With the DCE socket protocol it should be largely straightforward. But distributing the DCErouter itself would offer lots of reliability benefits, even if scalability isn't as necessary with the evident high performance of the router in its typical (residential) use cases. The hard part would be getting a DCErouter cluster to survive failure of one node (or subcluster) dedicated to, say, video or lights, decoupled from those supporting the intrusion alarms or telephony. Real security like that provided by independent dedicated systems, especially HW-only (despite their inconvenient compartmentalized messaging and UIs) can be closely approximated in LMCE only when the different processes run on independent hosts, and message signing is required between any individual SW objects, and IT intrusion detection is integrated into the whole system (and probably also pro IT remote management/sysadmin is available). All of which would still be less expensive and much more usable and flexible (and fully functional) than independent, dedicated, incompatible systems, even after adding all that expensive extra complexity to the currently fairly "naive" complexity that's perfectly adequate for most homes merely behind a firewall or two.

Lots of that could be best solved in the OS rather than at the app layer. I hope UHS, or some Linux project, is up to it. Because the LMCE project is going to be pretty busy before getting to those priorities.

[ddamron]

Sparkie,

Your disclaimer taken.

I think what you are seeing is more of 'advanced' storage configurations.. nas, san, etc.
To address your 'key ingredients', the simplest way is to simply add large drives to the core BEFORE setup.

The core will automatically configure, and use extra drives attached to it, without the hassle of SAN or NAS.

the key here is  the last sentence in you first paragraph: '..be attractive for a home user.'

adding drives to the core is the simplest method, and satisfies all your criteria.

Regards,

Dan

[1audio]

Having "used" Windoze Home server I think the only homes it will be OK in are those of network admins. Its really pretty clumsy to set up, has the user frendlyness of Vista, and it seems under some circumstances it can trash your data. http://gizmodo.com/338269/windows-home-server-corrupts-data-when-saving-from-certain-apps

An idea I had for higher end systems was a full "tandem" setup with two systems mirroring each other across a network. I would not be surprised if the architecture already exists in some Linux project or commercial implementation for banks and other critical applications. Rather than having individual failure points a full mirror will keep everything running until the power fails. And may be much easier to implement.

Having the system at the edge of the network is a mixed issue. If the user can handle configuring a router its possible to isolate it. But since Asterisk is inside dealing with NAT and ports is not for Joe Sixpack.