I don't think you should think of LinuxMCE as a program. Rather you should think of it as a unifying bridge architecture that draws together multiple and diverse applications and integrates them with a range of hardware, in order to present a plug and play experience to the user from the networked infrastructure upwards through individual items of functionality.
If you don't want it to be that (and the LinuxMCE network {or subnet} is a fundamental component of the system), then you may be better looking elsewhere.
That's more or less the conclusion I've come to. I already have a network. It feels wrong to duplicate functionality I already have, just to have to find hackish ways to avoid problems due to that duplication of functionality.
Don't kid yourself, though. You'll have to put in an awful lot of work on individual applications to get anything approaching the functionality of LinuxMCE.
Yes and no. At the moment, all I want is a good media center. In the future,say a year or two, I might to integrate some sensors from my alarm system. Home automation may arrive, but that's maybe fice years away. In other words, I don't need to duplicate all of it, just a few limited bits.
This thing here is a breathtaking achievement, almost without peer. It's flawed, and in early development stages, but don't let that blind you to what a stunning achievement it is.
Oh, I agree, and I think I've stated that several times. It was very impressive when it arrived, more or less out of the blue, and it's even more impressive now, and I certainly will keep a close eye on the project and forks of it. At the moment, though, it looks like my needs are a bit outside the scope of the project.
That's fine, we all have our views on things. It perhaps should occur to you that as the world becomes more and more integrated, and expects certain standards methods of connectivity, your system will run into increasing problems. It may be that you are correct and everyone else is wrong, but given that they will be able to do things, and you wont, it will be little comfort to you.
There will always be a possibility to have dedicated servers for each function, if nothing else because large networks demand it.
I was trying to help. I think you're profoundly mistaken in some of your beliefs (and yes, I am a professional working in the field). You have your view, and I wish you good luck.
Well, I appreciate the help a lot, you've given it a lot of time and effort. I don't share your views on some things, but, that's what comes with different experiences (and I'm also a professional). If one gets burned, one stays away from the thing that burned.
I must confess that I'm not sure MCE is for you. You have a lot of contraints which, it seems to me, you are not willing to flex on and which, as I see it, will result in you having a lot of trouble.
Agreed. I need something more flexible, and I'm prepared to give up some functionality for that.
MCE uses PXE booting. This is to ensure the software is distributed / mantained easily. You don't need to install stuff all over the place (core, MDs etc etc) You also don't require a local HDD in every MD, which reduces power consumption and cost.
Actually, I like that. Also, without a HD, you could put a huge passive cooler on the CPU and remove the fan from the PSU, making the device completely silent.
MCE needs to be your DHCP server. This is partly due to the PXE and partly due to the whole PnP concept. How else would the core "know" that there is a new device on the network? When a new device requests it's address, the core can use the MAC to determine if it's a fileserver / NAS thus needing scanning or a mobile device that needs the Orbiter image.
I see what you mean, but there are other solutions. PXE could definitely be set up on another DHCP server. As for differentiating machines, it could be done through a PXE boot which in an early step asks the core what it should be (although, I do realize that this would be a major rethink).
As for scanning servers, that's something I really, really do not want done automatically. As I've said, I have plenty of data. Even a simple ls of them takes 4-5 hours, so currently, I do it during the night, once a week, and then use that cached list. Faster, and easier on the disks.
MCE needs to be your router. This is because your ISP's router is probably set up as a DHCP server. Rather than having to go in and re-configure it (which isn't a very "non-tech" friendly requirement!), it's easier to make MCE dual-homed. The "public" NIC can play nicely with the ISP's router etc whilst the "private" NIC can do what it needs to do, including DHCP.
Makes sense for an ordinary user, but not for a technician like me. I don't even use any router from the ISP, just my own Smoothwalls. Hardware routers is something I have really bad experiences with, for instance, maxing out at 512 concurrent connections, which more or less makes P2P impossible.
Also, I see a distinct possibility that LinuxMCE might interfere with such things as VLANs and TOR. The thing that scares me is that once you leave a program to handle too much, you both become dependant on it and you lose control over it. It becomes a package deal, you get some good stuff, but some stuff does not work, and since it's a package deal, there's not much to do about it.
Now, looking at your setup, we have a completely different scenario. You are putting a lot of constraints on MCE which means that, IMHO, it won't actually do anything for you. You say yopu want 1 machine per function. Well, there's nothing to stop you installing MythTV on a box, Asterix on a box and so on. With all due respect, I think you are missing the point of MCE, which is why you are fighting it!
No, I'm not missing it, but I only need a subset of what it does. The rest just duplicates what I already have, and in a way that is somewhat awkward for me, or are things I don't even need. If I can get that subset (at the moment, a decent media center) without hassle, that's what I prefer.
I also think you are somewhat overly-inflexible about your current setup.
Well, I can't really afford much downtime, and I don't really have much time to spend on it. That means that I avoid tinkering.
I hear what you say about 1 public organisation, but there are millions of commercial networks out there that need and use managed switches very successfully.
In my opinion, that need is a need that can be avoided through proper planning and through placing the intelligence in computers instead of network devices.
MCE is a distributed, integrated system, it ISN'T a set of descrete computers on a network. It needs to work the way it does for reasons mentioned above. Stop thinking of it as taking over your network and think of it as a system on your network. If you go down the route of another network, you will have the MCE "system" existing on your current network with a single connection. The second network, whether provided with vLANs on your current structure, or with parallel cables, is part of your MCE system.
Sorry, but I can't decide to have my servers available on only one of two networks, physical or logical. They need to be accessible from both worlds. Forcing them to be on one and letting the communication to them go through the LinuxMCE core would provide a very nasty spof. The logical solution would be to just add another NIC to the servers allowing them to be on both networks, but there's no slots left in them, they are full of disk controllers.
Finally, you have already said you don't have the time to devote to helping develop MCE (although you seem happy to make use of other's work free of charge) yet you are prepared to put time and effort into re-engineering the system to fit your environment. This will be a huge task which will, I think, ultimately be doomed to failure.
Yep, I'm happy to use it free of charge, the same way I participate in other projects which are free of charge, as well as release some of my own stuff free of charge. I have every respect for the FOSS movement, but I have to pace myself and select some projects to support. I can't be everywhere.
Perhaps it will be futile. My plan is to make a quick test shot in a limited copy of my environment. Set up a separate network, modelled after the one I have, just with less machines. I can borrow one Smoothwall and internet connection for it, and use some spare desktops and laptops to simulate servers and desktops, as well as the core and the media players. Setting up that network can be done in a few hours, and then I can test it safely. If it works, good, if not, I'll check back later after a few versions.
Good luck, but I seriously doubt you will experience the success or have the pleasure of a system like the one seen in videos by people such as Thom....
Perhaps, but it's worth a try.
As I said, I really like LinuxMCE, otherwise, I would not go through this much trouble to make it work in my non-trivial network. Had it been lesser software, just suggesting that it must be run on 192.168.80.* would be enough for me to throw it out and not look back (by the way, with that limitation, how do you intend to handle networks with a need for more than 255 IP's, that's getting common in home networks now (or, possibly, me and my friends are not typical home networkers)).
