where do you record your shows to?

[big boi]

i recently switched over to recording shows directly to a network share using a UNC path. i watched one of the recordings last night and it seemed to have a few little "glitches" in playback. i'm fairly certain these were glitches in the actual recording, no just in playing back the file.

i have a newish pentium 4 system with onboard gigabit and 1GB of ram as my HTPC. in my server i have an 8 x 200GB raid5 array on a 3ware 7506-8. the server is a dul AMD 1900MP with 512MB of ram. i've got a gigabit card in the server and a gigabit switch between the 2 machines.

so my goal here is to find out how many of you record across the network, and have you ever had glitches in your recording. also, any suggestions are appreciated.

btw, i'm recording to DVD standard, and although i have 2 tuners only 1 show was being recorded.

[Blue]

The only glitch I had was rebooting the server while the wife was watching a recording. Now shows are recorded to a local drive on the HTPC.

When I was recording to the server, it was with a 100 Mb connection, and there were never any problems (other than the above). I wouldn't think your Gb connection would have any bandwidth issues.

What playback glitches are you having?

[Wheemer]

I really ment to vote for the third option, both local and network drives. Somehow my subconcious willed me to click the viking answer...

[big boi]

i watched the apprentice last night and it would seem to skip about 1 second of the show. like it was on a CD that was skipping or something is how i'd describe the effect. it was actually pretty bad in the apprentice.

after that i watched part of a northern exposure episode recorded the night before. there was a tiny tiny part right at the beginning where joel tunrs his head toward the camera and it just looked slightly blurred. barely noticeable, but it bothered me. i replayed the scene several times and it was the exact same effect each time so i know it was a glitch in recording the show, not a glitch from the playback. the defect was recorded into the file if you see what i mean. then i watched another 30 mins of the episode and there were no issues at all.

[Opus4]

You should be able to record over that network OK... my first thought when I hear about recording/playback glitches is to ask if the partition is using a 64K cluster size. I don't know whether/how the raid5 setup would affect it, though.

- Andy

[big boi]

would that be the default cluster size? is there a way to check on this? the array uses different stripe sizes (i believe mine is set to 64K) but this is totally seperate from what you're talking about and only the raid card is aware of the stripe size.

[big boi]

okay, well i see the default cluster size is 4KB so i'll assume that's what my array is formatted as. guess i'll be moving all the data off my array tonight (should take all night). then reformat the array as per the manual (yes, RTFM jacka$$ ) then copy all my data back. what is the saying? an ounce of prevention is worth a pound of cure?

thanks

[salsbst]

I record to a local drive on the 'server'. I marked as 'local drive on the HTPC'. There's very little TV watching happening directly on the server, though. Clients in the TV room and bedroom connect to the server in the study.

I'm presently setting up a new file server that should have the bandwidth for several simultaneous HD streams. At some point, I will record from the server HTPC to the file server over a dedicated 1Gbps connection. For now, though, there are other, higher priority network reorganization tasks and hard disk practicalities that are preventing me from switching immediately.

[bartley9]

I formatted the 250GB drives on my network server using default cluster size. No problems using SageTV and one SageTV client(Not tried more).

100GB ethernet network.

[big boi]

i don't really understand what you're saying. you are running a sage server which records to a drive that is in that computer? and then eventually you'll record across the network? that's what it sounds like. correct me if i'm wrong.

Originally Posted by salsbst

I record to a local drive on the 'server'. I marked as 'local drive on the HTPC'. There's very little TV watching happening directly on the server, though. Clients in the TV room and bedroom connect to the server in the study.

I'm presently setting up a new file server that should have the bandwidth for several simultaneous HD streams. At some point, I will record from the server HTPC to the file server over a dedicated 1Gbps connection. For now, though, there are other, higher priority network reorganization tasks and hard disk practicalities that are preventing me from switching immediately.

well i have to say i hope the cluster size is the problem because if it's not i'm going to be doing a lot of work for nothing AND i'll still have this problem to fix

I formatted the 250GB drives on my network server using default cluster size. No problems using SageTV and one SageTV client(Not tried more).

[Opus4]

Originally Posted by bartley9

I formatted the 250GB drives on my network server using default cluster size. No problems using SageTV and one SageTV client(Not tried more).

You are using a 4K cluster size, then? How long have you been using that drive for SageTV recordings? And, what recording rate are you using? I won't say there is a 100% probability that you will eventually run into recording/playback glitches, but... when the drive starts getting more fragmented, your recording rate is high enough, and/or you are recording/playing multiple shows at once, please remember the small cluster size before asking us what is causing poor recordings or jerky playback.

In other words: I don't know that _everyone_ with a 4K cluster size will run into problems, but it is certainly much more likely than for someone using a 64K cluster size. And, I do not know if it has the same effect on a raid5 array.

- Andy

[big boi]

well considering i've never heard of the 64KB cluster recommendation i would say i've been using this for 8 months now. however, for most of that time i was recording to a drive local to the HTPC and it was replaced once. so the local drive was used for 4 months, i swapped in a bigger drive and used that for 4 months. perhaps fragmentation never became an issue.

with the network recording it's been in use for under a week and the problems were noticeable on basically the first recording i watched. so, i'm going to bet your diagnosis is the right one and fix it. i'm hoping you're right because in the scheme of things it's not that bad to fix and i'll have everything the way i want it within a day or 2.

Originally Posted by Opus4

You are using a 4K cluster size, then? How long have you been using that drive for SageTV recordings? And, what recording rate are you using? I won't say there is a 100% probability that you will eventually run into recording/playback glitches, but... when the drive starts getting more fragmented, your recording rate is high enough, and/or you are recording/playing multiple shows at once, please remember the small cluster size before asking us what is causing poor recordings or jerky playback.

In other words: I don't know that _everyone_ with a 4K cluster size will run into problems, but it is certainly much more likely than for someone using a 64K cluster size. And, I do not know if it has the same effect on a raid5 array.

- Andy

as for how this relates to the fact this is a raid5 array, all of this information should still apply. the cluster size is important to windows and fragmentation and seeking will still be affected by this. with a hardware raid adapter windows is being presented with a single massive hard drive at the low, hardware level. it does not know anything that is going on beneath that. so NTFS is going to function just as it normally would if this was actually a single 1TB hard drive.

the only thing is that you now also have to think about the way the raid card is seeing the disks. the raid card is writing data to the disks in chunks you select when creating the array. for multimedia storage 64KB stripes is a good size for basically the same performance reasons listed for using 64KB clusters. with a larger stripe size like this the raid card breaks a massive file into fewer pieces allowing for less parity calculation work, and it has to break these files down less.

so anyway, i'll make some changes and report back how it goes. in fact, the first thing i'll do when i get home is simply check the fragmentation of the array and see if it's in bad shape.

[salsbst]

Originally Posted by big boi

i don't really understand what you're saying. you are running a sage server which records to a drive that is in that computer? and then eventually you'll record across the network? that's what it sounds like. correct me if i'm wrong.

I think you got it...

For now, I record to drives that are local to the HTPC server.

Within a few months, I will record to a file server instead of local drives.

Now and in the foreseeable future, the client HTPCs will get their video streamed by the HTPC server.

Sorry if this is not clear. I don't know how else to say it .

[big boi]

i guess it was clear enough. it was the use of the term "server" when referencing the file server v. when referencing the sagetv server that through me off.

Originally Posted by salsbst

I think you got it...

For now, I record to drives that are local to the HTPC server.

Within a few months, I will record to a file server instead of local drives.

Now and in the foreseeable future, the client HTPCs will get their video streamed by the HTPC server.

Sorry if this is not clear. I don't know how else to say it .

[big boi]

in the process of moving things off the array so i can reformat it, but in the meantime here were some disk stats.
free space: 32%
total fragmentation: 11%
file fragmentation: 22%
defrag suggested by windows?: yes

[dvd_maniac]

in the process of moving things off the array so i can reformat it

Couldn't you have used an on-the-fly utility like Partition Magic?

[bartley9]

I have a dual P350 file server which has 5x250GB disks, all formatted at default blksize running WIN-NT Server with 512MB RAM.

Sage runs on a little VIA-EPIA-M1000 and records across the 100MB network to the first disk in the server. I have defined the other 4 disks to sage as UNC import directories.

On the server I manually copy shows I want to keep from the "Live" disk to the other 4 as needed. As I don't delete shows from these disks and never copy from more than one source location at a time there should be zero fragmentation.

I have had the first disk since September '04 and have not noticed any glitches. I only have one tuner. Perhaps WIN-NT does read ahead processing?

[big boi]

Originally Posted by dvd_maniac

Couldn't you have used an on-the-fly utility like Partition Magic?

i'd rather start fresh. besides, i cleaned up about 100GB of crap i no longer needed or had duplicates of.

since read ahead processing has been mentioned- what should my cache on the raid controller be set to? enabled or disabled? i am going with 64K stripes for the raid5 array and 64K cluster size. files will be moved back tonight and i'll post recording results later this week.

[big boi]

well the first recording to the fresh partition was done last night. i watched about 3/4 of the northern exposure episode and it did not skip at all. i'll be sure to keep an eye on things and let y'all know if the problem recurs though.

this brings up another question though. so for those of you recording across the network, how many shows are you able to record at once? i have the hauppauge 500mce and wonder if a dual recording situation might give me trouble. what if i add another card?

i was thinking about this on the way into work this morning and given that i have about 700MB of free ram in my computer it'd be nice if sage could effectively make use of that to buffer the network recordings. you ought to be able to squeeze like 15 minutes of recording into that RAM. i dunno, just something i was thinking about. i'm not critiquing though since i'm not a programmer and have no idea what that would involve.

[fieldeffect]

I don't record across the network, but a 100Mb network can realistically transfer about 10MB a second of actual data under most circumstances, which is about 36GB per hour.

Considering that a best quality Sage recording is around 3GB per hour the network shouldn't cause any problems.

Having said that, if I was going to use a network that is used for other potentially high bandwidth traffic i'd probably run it through a Linux box and do some packet shaping to guarantee bandwidth on the Sage ports.

Or the cheaper way would be to throw an extra NIC in both the Sage machine and the server and have a dedicated network between the two.

Nick.