How does Sage handle mounted drives?

[blade]

I know Sage looks for the drive with the most free space and places new recordings there.

Let's say I have a drive set up as D.

I create the recording directory D:/Recordings. Next I mount another physical drive as D:/Recordings2 and another as D:/Recordings3. Will Sage be able to determine the free space for each directory and allocate the recordings accordingly or will the fact they all share the same drive letter cause problems?

Also is anyone using mounted drives in Windows and are there any potential drawbacks?

[blade]

I guess no one is using mounted drives with Sage.

[ke6guj]

Nope, don't use mounted drives.

However, I did once specify 2 recording directories on the same HD once, and IIRC, it showed me as having twice my actual free space. I didn't run it very long like that, so I don't know how it would have allocated the recordings.

edit:
I just tested it, and it seemed to work fine. It looks like Sage was able to look into the mounted drived and see the correct amount of free space in that folder, so I don't see why it couldn't properly allocate the recordings.

Also is anyone using mounted drives in Windows and are there any potential drawbacks?

My main question is, why? Is there a benefit I'm not seeing? Perhaps mounting a UNC drive to a local folder instead of just mapping to a drive letter.

[blade]

I've noticed Sage doing that as well, which was one of the reasons I was wondering if it would handle it properly. It would be a lot cleaner to have one drive letter for all of my recording drives.

Currently I have 3 drive letters on the server, 3 seperate UNC paths so the client can use comskip, then dirmon is setup with 3 seperate paths to check. It would be a lot cleaner to have 1 drive letter on the server, 1 mapped drive on the client, and 1 path for dirmon.

[ke6guj]

Ok, that makes sense.

Your last post came as I was editing my last, but I did some testing and I did see that if you list 2 real directories off the same HD, then Sage doubles the free space, but if 1 directory is a mounted drive,then it did correctly calculate the free space.

[BobPhoenix]

Maybe I'm not understanding correctly but what I have tried is this:

C:\RecordingFolders with subdirectories that are mapped to drives D and E. The names of the folders are DriveD and DriveE. Then I added the Recorded TV directories that are on each drive to SageTV as c:\RecordingFolders\DriveD\Recorded TV and c:\RecordingFolders\DriveE\Recorded TV and the space reported free is correct. But if you use something like \\mediacenter\recordingfolders\drived\recorded tv and \\mediacenter\recordingfolders\drivee\recorded tv the space available is incorrect. So looks like it might work with direct drive mapping but not with UNC mappiing. Which is probably to be expected.

BobP.

[ke6guj]

Somethings definitely weird here. I saw the same behavior as Bob did with UNC mappings to mapped drives.

I think we will need some offical info on how Sage calculates free space. When does Sage look at the amount of free space in the folder, and when does it look at the amount of free space in the drive? Because, it seems to use both methods.

And it doesn't seem to log the calculations in the log files, it only shows me the available space for my real drive. It does not show the available space of the mapped drive, or the "doubled" space if a second folder on the same drive is added, even though the doubled space shows up as "available video diskspace". And I was able to get Sage to record to the mapped drive if I reduced the amount of space that Sage could use on the real drive.

Jack.

[blade]

Thanks for testing it ke6guj. I've been using mapped drives which gets to be a mess on multiple machines each requiring multiple mapped drives.

I tried changing to UNC paths to clean things up a bit; however, it adds overhead to the server even when recording locally. My server is very low powered so I ended up switching back to mapped to avoid the wasted cpu cycles. I'm assuming mounting drives won't add the extra overhead that using UNC paths does. I may be wrong because I'm just guessing.


BobP are you talking about mapping the drives or mounting them? What I'm talking about is instead of having the drives on the server (where they're installed) show up as D, E and F only D will show up. Drives E and F will be mounted as directories under the D drive.

I want to have 3 recording directories under D, one for the physical D drive and then 1 for each of the physical drives that are mounted under D. Then I'll map D on the client and I'll have Dirmon scan the D drive and it's subdirectories. This way I can use mapped drives and regardless of how many recording drives are in the server I will only need to setup 1 mapped drive on the clients.

[GKusnick]

A simpler option might be to use dynamic disks, which let you create a single logical volume that spans multiple physical drives. Then you could have just one recording directory, which you could grow as needed by adding more drives.

Removing drives is a bit trickier. Once you've added a drive to a spanned volume, you have to copy everything off that volume and delete the whole volume in order to free up the individual drives.

Still, if all you need is a way to pool the disk space from multiple physical drives under one drive letter without having to mount them as subdirectories, spanned volumes might be worth a look.

[ke6guj]

IIRC, with a spanned volume, there is no redundancy. If one drive dies, you lose the entire span, sorta like RAID 0.

[blade]

I had thought about using spanned volumes when I first setup my server, but didn't like the thought of one drive dying killing everything.

[BobPhoenix]

Originally Posted by blade

BobP are you talking about mapping the drives or mounting them? What I'm talking about is instead of having the drives on the server (where they're installed) show up as D, E and F only D will show up. Drives E and F will be mounted as directories under the D drive.

I want to have 3 recording directories under D, one for the physical D drive and then 1 for each of the physical drives that are mounted under D. Then I'll map D on the client and I'll have Dirmon scan the D drive and it's subdirectories. This way I can use mapped drives and regardless of how many recording drives are in the server I will only need to setup 1 mapped drive on the clients.

What you describe is exactly what I was talking about. Drive C is the only drive letter and the Drive that was D is mapped to folder DriveD and the Drive that was E is mapped to folder DriveE. When I did that the size of the used and free space was reported the same as when they were separate drives but when I used a UNC path shared folder that contained the mapped drive folders DriveD and DriveE then the reported space was what I had free on my C drive not the free space on D and E added together. So that lead me to believe this would work with mapped drives but not with a UNC share.

BobP.

[blade]

Originally Posted by BobPhoenix

What you describe is exactly what I was talking about. Drive C is the only drive letter and the Drive that was D is mapped to folder DriveD and the Drive that was E is mapped to folder DriveE. When I did that the size of the used and free space was reported the same as when they were separate drives but when I used a UNC path shared folder that contained the mapped drive folders DriveD and DriveE then the reported space was what I had free on my C drive not the free space on D and E added together. So that lead me to believe this would work with mapped drives but not with a UNC share.

BobP.

Ok, just wanted to make sure we were on the same page.

[heffe2001]

Originally Posted by BobPhoenix

What you describe is exactly what I was talking about. Drive C is the only drive letter and the Drive that was D is mapped to folder DriveD and the Drive that was E is mapped to folder DriveE. When I did that the size of the used and free space was reported the same as when they were separate drives but when I used a UNC path shared folder that contained the mapped drive folders DriveD and DriveE then the reported space was what I had free on my C drive not the free space on D and E added together. So that lead me to believe this would work with mapped drives but not with a UNC share.

BobP.

The remote machine has no idea that the drives are mapped to folders, so it pulls the total space from the drive the share is located on. Don't know any other way around that.

Raid is the way to go if you want ONE share. Just use Raid5 and you also don't have the problems with drive-loss (but do loose a considerable amount of space).

Or better still, drop your PC clients, and stick MVP's on there, then no drive maps to make (but then again, no HD support YET).

[BobPhoenix]

Originally Posted by heffe2001

The remote machine has no idea that the drives are mapped to folders, so it pulls the total space from the drive the share is located on. Don't know any other way around that.

That's what I was figuring as well.

[dshields]

Originally Posted by heffe2001

The remote machine has no idea that the drives are mapped to folders, so it pulls the total space from the drive the share is located on.

I can confirm this. Here's my post about it:
http://forums.sagetv.com/forums/show...1&postcount=13