RocketDivision.Com

lackmaster · Joined: 23 Jan 2006 Posts: 6

Hello,

I have 2 Win2003 servers each is using MS iSCSI initiator to mount a 800GB F: drive on an EMC CX300i and an 800GB G: image file drive on a Starwind v2.6.4 Win2003 SATA300 storage server.

I am getting numerous failures in copying files from F: to G:. Most of the files copied fine. I can copy the failed files from F: to other places, just not to G:. The following is an example of the error I receive:

2006/01/23 13:29:40 ERROR 1117 (0x0000045D) Copying File F:\directory\subdirectory\filename.img
The request could not be performed because of an I/O device error.

Has anyone seen this before?

valery (staff) · Joined: 29 Jun 2004 Posts: 280

lackmaster · Joined: 23 Jan 2006 Posts: 6

No, none of the iSCSI targets are shared. I have 2 different ones in Starwind for 2 different servers. Same for the EMC targets.

My log file is 1.2 Gig large, I can't even open it. I will only send from the beginning of the file up through where the errors begin.

anton (staff) · Joined: 18 Jun 2004 Posts: 1128 Location: Ukraine

OK, strip it down and let us have one of them for analyze. Thanks!

valery (staff) · Joined: 29 Jun 2004 Posts: 280

lackmaster · Joined: 23 Jan 2006 Posts: 6

It looks like there is some kind of limit as to the number of files that some directories can hold on my Starwind image files.

I have a directory 5 levels deep with 6 files in it. If I delete 1 file from that directory, then I am able to save any 1 file back to it, but only one file, if try to save another file to it then I get the I/O error message.

If I delete 2 files from it, then I am able to save any 2 files back to it.

However, at the directory above that one, I can save as many files as I want.

Specs from one of my Starwind drives:
849.61 GB total space
1,772,286 Files, 4,329 Folders = 201 GB

My other Starwind drive is 867.87 GB total size and has about 2.5 times as many files, and about 6000 folders.

There is nothing further in my event logs except for the error 1117. It's not a resource problem either.

If you don't have a fix for this then I will have to use a different product.

valery (staff) · Joined: 29 Jun 2004 Posts: 280

Will,

Is the ImageFile a flat image or a sparse one?
If it's sparse, is there free space on the partition where the ImageFile located?

We're trying to reproduce the problem in our test labs.
If there is a bug in our code we'll fix it ASAP.

Thank you.

lackmaster · Joined: 23 Jan 2006 Posts: 6

I guess the image files are sparce since I did not choose "flat" when I created them.

I have several TB of free space.

What is the difference between flat and sparce?

valery (staff) · Joined: 29 Jun 2004 Posts: 280

lackmaster · Joined: 23 Jan 2006 Posts: 6

I will try creating a flat file image and use it to see if I still have problems. This will take a while.

valery (staff) · Joined: 29 Jun 2004 Posts: 280

anton (staff) · Joined: 18 Jun 2004 Posts: 1128 Location: Ukraine

Generally speaking WriteFile shoud not fail with "no resources" on sparse files as well - task to our QA to work deeper. However there could be issues with configuration of the target on the server machine - like having 3 TB of free disk space and say 100 MB set for "disk quota". And so on...
_________________
Regards,
Anton Kolomyeytsev

Rocket Division Software

lackmaster · Joined: 23 Jan 2006 Posts: 6

I re-created the imagefile:

900 GB
Flat file
Asynchronous

I relaunched the robocopy job that replicates the F: drive to the G: drive, and have NOT seen the same errors I was seeing before.

It has completely replicated 547 GB of 50 KB files (which is an enourmous amount of files) without any errors. As it was before, it started throwing errors around 250 GB.

I will now do the same for the other server and see if I get the same good results.

valery (staff) · Joined: 29 Jun 2004 Posts: 280