Seperate "tuner box" from main server?

[kameleon]

I have a decently nice main media pc that I have 2 Hauppauge PVR-500's in for tuning regular analog cable tv. I have a spare server license that I am not using currently and want to move those tuners into a separate box that does nothing but tune tv for my 3 clients.

So my question is this: My media and all will still be on my main media pc but will I be able to move the tuners into a separate box and still have my clients connect to it for live tv AND my main server for the stored media? Also I would still want the main server to handle the transcoding and commercial removal since it has a nice dual core 3Ghz cpu with 8GB ram vs the tuner box that only has a single core 2Ghz with 1GB ram.

Is this possible? Thanks in advance.

[GKusnick]

What you're talking about is called a network encoder and is described in one of the appendices of the user manual (sorry, I don't have it in front of me at the moment).

[Fuzzy]

I would, however, advise keeping the storage with the tuners, and using the extra license as a client instead. Recording over the network is a worst case scenario, because the most critical function of sage is the reliable recording. Requiring two systems to be up and operational for the recording to take place is far from ideal.

[will]

Originally Posted by Fuzzy

Recording over the network is a worst case scenario, because the most critical function of sage is the reliable recording. Requiring two systems to be up and operational for the recording to take place is far from ideal.

Do you achieve the same quality using SageTV as a network encoder? Are there any downsides to this besides doubling your chances of system failures?

[Fuzzy]

The quality is exactly the same. The encoding is done either by the capture card, in the case of analog capture, or by the broadcaster, in the case of digital capture. Moving it to a different system won't affect it in any way.

[sflamm]

Today I do this very reliably with the Ceton in a dedicated small form factor PC (shuttle). No issues at all.

The alternative is to use the new Silicondust Prime in the exact same role / configuration - which based on my previous experience with HDHRs is I'm sure also very reliable.

[kameleon]

That is exactly what I was looking for! I had seen/read about it way back when but totally lost it when I needed it most. Hopefully I will have all this done by this weekend. I just wish I had a way to move the storage into the same enclosure as the tuners but the only way I can accomplish that is by keeping them in the existing system since it has 6 sata ports. The spare motherboard I have for the tuners only has 2 IDE ports and I don't have any good big drives to add.

[kbyrd]

Originally Posted by Fuzzy

I would, however, advise keeping the storage with the tuners, and using the extra license as a client instead. Recording over the network is a worst case scenario, because the most critical function of sage is the reliable recording. Requiring two systems to be up and operational for the recording to take place is far from ideal.

++ to this.

Keep the recording path simple. If you drop data somewhere between signal and hard disk, it's gone. I've even considered putting the HDHR directly connected to my sage server on it's own network card, but since I haven't had trouble with my current configuration, I'm leaving it alone.

[will]

Originally Posted by kbyrd

Keep the recording path simple. If you drop data somewhere between signal and hard disk, it's gone.

How is this any different than using an HDHR; which users seem to have a good experience with?

I would love to be able to hookup my tuners to a tiny PC running an atom CPU and then virtualize my storage arrays and servers (including my main Sage server). I'm reconsidering a lot of my setup as I am in the process of building out my main blu-ray/dvd video storage array from a RAID 5 6 tb (four 2 tb discs) to a RAID 6 30 tb (12 3 tb discs plus a 3 tb hot-swap).

Clouds be damned

[Fuzzy]

There's no reason those couldnt' all reside in the same system (and, in my mind, they SHOULD reside in the same system). The biggest problem with virtualizing everything is that drive performance is still pretty poor between different VM's, because it is going through the Virtual networking layer, which for some reason still seems limited in most cases to Gbps speeds, which, once you add in networking overhead, is a LOT slower than direct drive access. So, for instance, running watching a livetv broadcast from an or client connected to your virtualized multi-headed beast would involve the data flowing from the tuner card (Through the VM's passthrough feature), to the sagetv server's VM, from there, written over the virtual network to the storage VM's raid array (software raid perhaps?), then re-read from there, back to the virtual network to the sagetv server VM over the same virtual network, then from there over the real network to the client for playback. That is a LOT of overload for just a single livetv viewing session. Throw a few of those in the mix, and you will start to run into problems that storage local to the tuners would have avoided.

To me, the only reason to separate them in VM's as you propose, would be if you wanted to use an advanced software RAID system (like unraid, for instance) for your storage system, and use the windows sagetv server. I, personally, don't feel tv recordings warrant that kind of real-time protection. You can use a windows compatible snapshot parity system (like FlexRAID, or even PAR2 files) if you want to make them recoverable from a drive failure, and not sacrifice the performance.

[kbyrd]

Originally Posted by will

How is this any different than using an HDHR; which users seem to have a good experience with?

My system is a HDHR connected to a sage server with internal disks. The HDHR is an appliance. It's on or off. It's probably more likely to die than a PCIe card. But I'm not worried about complete failure so much as I am glitches during recording. In an ideal world I would use a PCIe card to simplify things, but the HDHR gets me two things:
- Sage Server runs Linux and I wanted to avoid tuner card driver crap.

- I can place the HDHR "closer" to the income cable source than I can the full Sage Server. I can run Ethernet cable 200 feet between the HDHR and the Sage Server without worrying about a marginal incoming QAM signal attenuating further.

To answer your question (finally), I've done this both ways and while I haven't explored it rigorously, I have always a better experience with the storage local to the Sage server doing the writing. This was true whether I was using an HDHR or a PCIe tuner. Assuming modern GigE networking gear and good wiring, the network isn't the problem. When you write to a NAS, there's a lot of bits in the stack that just aren't there when reading from an HDHR. Maybe I just didn't find the right combination of tuning, but I've never been happy with large streaming writes over NFS of CIFS even with perfect network connection.

[kbyrd]

Originally Posted by Fuzzy

There's no reason those couldnt' all reside in the same system (and, in my mind, they SHOULD reside in the same system). The biggest problem with virtualizing everything is that drive performance is still pretty poor between different VM's, because it is going through the Virtual networking layer, which for some reason still seems limited in most cases to Gbps speeds, which, once you add in networking overhead, is a LOT slower than direct drive access. So, for instance, running watching a livetv broadcast from an or client connected to your virtualized multi-headed beast would involve the data flowing from the tuner card (Through the VM's passthrough feature), to the sagetv server's VM, from there, written over the virtual network to the storage VM's raid array (software raid perhaps?), then re-read from there, back to the virtual network to the sagetv server VM over the same virtual network, then from there over the real network to the client for playback. That is a LOT of overload for just a single livetv viewing session. Throw a few of those in the mix, and you will start to run into problems that storage local to the tuners would have avoided.

To me, the only reason to separate them in VM's as you propose, would be if you wanted to use an advanced software RAID system (like unraid, for instance) for your storage system, and use the windows sagetv server. I, personally, don't feel tv recordings warrant that kind of real-time protection. You can use a windows compatible snapshot parity system (like FlexRAID, or even PAR2 files) if you want to make them recoverable from a drive failure, and not sacrifice the performance.

A couple of things:

For current versions of vSphere (and probably Xen) VM-to-VM networking is limited by CPU cycles. If you're seeing an artificial limit, something is wrong.

I have spent a lot of time thinking about how available I want my recordings. I've tried things like FlexRAID but I always return to Linux md. I agree that I don't mind if I actually lose the recordings. But real RAID gets me two things: - speed. Six disks is fast enough that I don't worry about ripping a blu-ray watching another blu-ray rip, while recording two HD QAM streams, while also doing a backup. This is probably overkill, but I had trouble with two recordings while watching a third from a single disk. It just wouldn't seek fast enough I think.

- While I wouldn't actually care if several of the recordings disappeared right now, what I don't want to happen is for the entire set of recordings to go away while I'm out of town and my wife just wants to watch TV.

[Fuzzy]

The multi-record and watch scanario is best served with separate recording paths, and sage set to performance mode, than it is running a RAID array. The problem is the constant seeking between recording files, which is eliminated with multiple separate drives. I used to run a raid-5 array and had a LOT more drive delay than I get now with split drives (now that sage has the performance/bandwidth mode). Any single HD is more than capable of a couple HD streams, if the seeking is kept to a minimum, so the sequential access bandwidth benefits of RAID are unnecessary.

In truth, I might still be running the RAID-5, if win7 supported it like my older hacked up xp did... but I'm glad I was moved away from it. I have never had a drive performance problem on my win7 server, with 5 separate recording drives, and 5 tuners (2 R-5000, 3 ATSC).

[kameleon]

I hope someone has some pointers. I have setup both machines like I wanted:
winxp with 2 PVR-500's
win7 setup as main sage server

Using the "network encoder" portion of the appendix I ran sagetv on the winxp machine with the line

"C:\Program Files\SageTV\SageTV\SageTV.exe" enable_encoding_server=true

and the main win7 sage server with:

"C:\Program Files (x86)\SageTV\SageTV\SageTV.exe" network_encoder_discovery=true

I have the recording location set properly to the win7 shared drive. I also have both machines firewalls OFF for testing purposes. I am unable to get the win7 machine to see the winxp tuners though. What could I be doing wrong? I really would like for this to work but if I can't get it to work like this I will simply move the tuners back into the win7 machine and drop it back to 4GB ram. Any ideas? Thanks in advance!

[Fuzzy]

Originally Posted by kameleon

I hope someone has some pointers. I have setup both machines like I wanted:
winxp with 2 PVR-500's
win7 setup as main sage server

Using the "network encoder" portion of the appendix I ran sagetv on the winxp machine with the line

"C:\Program Files\SageTV\SageTV\SageTV.exe" enable_encoding_server=true

and the main win7 sage server with:

"C:\Program Files (x86)\SageTV\SageTV\SageTV.exe" network_encoder_discovery=true

I have the recording location set properly to the win7 shared drive. I also have both machines firewalls OFF for testing purposes. I am unable to get the win7 machine to see the winxp tuners though. What could I be doing wrong? I really would like for this to work but if I can't get it to work like this I will simply move the tuners back into the win7 machine and drop it back to 4GB ram. Any ideas? Thanks in advance!

That's not how you do it. You need to edit/add those lines to each install's sage.properties file (its a text file in the sagetv folder). then just run sagetv.exe.

[kameleon]

Originally Posted by Fuzzy

That's not how you do it. You need to edit/add those lines to each install's sage.properties file (its a text file in the sagetv folder). then just run sagetv.exe.

Ahhh I thought that was the "old way" from everything I read. The way I was trying is the one in Appendix G of the manual that comes with v7. I will give the "old way" a shot and see how it goes and report back. Thanks!

[Fuzzy]

Appendix G says nothing about launching sage with any command line parameters, other than to say that you can use the -client parameter to launch a second instance on the encoder machine to still use it as a client.