Scandisk notification on every reboot of the target server.

[Frater Kork]

Hello Guys.
Wonderful apps StarWind/StarPort!
I have strarted messing with them for my ATA server environment.
What worries me is that the Machine running StarWind always wants to perform a Scandisk on all exported disks every time I restart it.
Usually it finds no errors, but it implies that the NTSC partitions are out of synch in some way I guess.

I do not access the disks locally, and the server is doing pretty much nothing except holding a lot of disks and a network card..
Presumably this has something to do with that server 2003 is doing background indexing on them while they are being accessed...

I am running a plain vanilla Windows Server 2003 with all patches and the latest StarWind from your site as of this date. The disks are formatted with NTFS and Basic partitions.

Cheers!

[anton (staff)]

If you're sharing your devices as SPTI ones I'd recommend not doing this at all. At least with the file system Windows 2003 (or whatever you use on target machine) recognizes (EXT2 or XFS or whatever Linux or Mac has on initiators are OK). See, Windows still trys to write something to disk even if it's opened in exclusive mode by SPTI application. Sick!

I would recommend you either exporting storage as hard disk partitions (using non-zero offsets so local file system driver would not pick them up) or as image files. Rock solid way. SPTI works great for tapes, CD/DVD-RWs etc. For hard disks with "alien" file systems. But not NTFS...

Hello Guys.
Wonderful apps StarWind/StarPort!
I have strarted messing with them for my ATA server environment.
What worries me is that the Machine running StarWind always wants to perform a Scandisk on all exported disks every time I restart it.
Usually it finds no errors, but it implies that the NTSC partitions are out of synch in some way I guess.

I do not access the disks locally, and the server is doing pretty much nothing except holding a lot of disks and a network card..
Presumably this has something to do with that server 2003 is doing background indexing on them while they are being accessed...

I am running a plain vanilla Windows Server 2003 with all patches and the latest StarWind from your site as of this date. The disks are formatted with NTFS and Basic partitions.

Cheers!

[Frater Kork]

Hmm, too bad, I'm not much of a linux person.
It would have been soo easy to manage this solution if windows just kept its tentacles off the disks that dont contain any system or related files.
Using NTFS it's five minutes work to move a disk from one machine to another and keep the data intact.
If I move to image files I get kind of stuck in one software solution.

What is this non-zero offset partitioning You mention?

[Alex (staff)]

Hi,

I saw this check a couple of times. I will take care of this.

[Frater Kork]

In an attempt to stop windows on the target machine from messing with the shared data disks I removed the System Volume Information as well as Recycler folders and then went into disk manager and removed the drive letter from one of the disks.
This should render the disk logically inaccessible to the local windows installation so it wont mess with the disk by adding new sysvol and recycler folders etc, while keeping the NTFS partition with data intact on it.

Surprisingly this locks up the Initiator machine a short moment after adding the physical disk in StarPort, it hangs to the degree that I need to powercycle it.
When it launces the StarPort service it hangs again when trying to mount the persistent drive, so the only way out is to remove the network connection to the Target server and then remove the disk connection...
Connecting a second disk that I had not removed the Mount Point on the same target machine worked fine however (still checkdisk on that one however)

Dont know if this would have removed the initial corruption problem, but it was an idea anyway

Target: Server 2003
Initiator: 2000 Pro

[Alex (staff)]

Surprisingly this locks up the Initiator machine a short moment after adding the physical disk in StarPort, it hangs to the degree that I need to powercycle it.

Strange. I'll check this. How did you share the disk - as SPTI device or using ImageFile functionality?

[Frater Kork]

I am trying with SPTI because I dont want to bind my data in a proprietary format.

[Val (staff)]

I am trying with SPTI because I dont want to bind my data in a proprietary format.

Hi Frater,

We don't recommend using the StarWind's SPTI plugin for sharing hard drives.
It seems there are some issues in current StarWind SPTI implementation that can damage integrity of a hard drive remotely accessed with it.

Please use the ImageFile plugin instead to share separate partitions of your disc.

We'are working on the issue...

Thank you for your feedback.

[Frater Kork]

Cool

Queuing the access calls to the disk via the same channels that the hosting windows application uses may well be the way for getting SPTI to work without corrupting the data. It will probably be a bit slower that the current implementation, but what the hell...
The current SPTI is just as multi user as the one you warned cluster users from

By the way, why do You not support Dynamic partitions on the disks?

[Alex (staff)]

Queuing the access calls to the disk via the same channels that the hosting windows application uses may well be the way for getting SPTI to work without corrupting the data. It will probably be a bit slower that the current implementation, but what the hell...
The current SPTI is just as multi user as the one you warned cluster users from

SPTI is a back door actually.

By the way, why do You not support Dynamic partitions on the disks?

You cannot share dynamic disks via SPTI correctly. Host and client will fight for it - both will try to write own unique signature to disk's LMD and one of them will see the disk as offline.

You can use dynamic disks if they are managed as ImageFile devices on the StarWind side. Anyway each time when you will connect from different client computer you will need to reactivate the disk. (Because of the same reason as above).

AFAIK support of automatic dynamic disks reactivation is ready but not yet added to StarPort. It will be added very soon I suppose.

[anton (staff)]

Yup. SAN file system is what should be used on the disk shared AS IS (via SPTI). Something what guys from www.dataplow.com offer. We're testing their stuff right now. Forks fine. You can contact them directly for evaluation and purchasing of their product. Much better idea then use NTFS

Yup. Dynamic disks reactivation on client side is done and is expected to be added in next major update (together with German localization if you care). Hold on.

Queuing the access calls to the disk via the same channels that the hosting windows application uses may well be the way for getting SPTI to work without corrupting the data. It will probably be a bit slower that the current implementation, but what the hell...
The current SPTI is just as multi user as the one you warned cluster users from

SPTI is a back door actually.

By the way, why do You not support Dynamic partitions on the disks?

You cannot share dynamic disks via SPTI correctly. Host and client will fight for it - both will try to write own unique signature to disk's LMD and one of them will see the disk as offline.

You can use dynamic disks if they are managed as ImageFile devices on the StarWind side. Anyway each time when you will connect from different client computer you will need to reactivate the disk. (Because of the same reason as above).

AFAIK support of automatic dynamic disks reactivation is ready but not yet added to StarPort. It will be added very soon I suppose.