short names for what ?

I guess I didn't word it correctly. I need to disable auto short name generation for file names.

hmm..it must be something in Celerra i did not know about ..or you are confusing Celerra with Centera ?

Are you referring to Windows generating 8.3 filenames? As in "Windows Stuff.doc" turning into "WINDOW~1.DOC"?

The Windows whort names with the ~1 at teh end would be fine. But, what's happening is the names are being randomly generated.

Example:

U15070041.001 is changed to C862CC84.001.

The problem with this is, these are scanned files which are separated by the first letter. Cxxxx files go to a tif splitter program. But since some files that start off as Uxxxx files are being changed to files starting with C, the files are ending up in the wrong folders.

what type are your clients ?

DOS or some old Windows that doesnt speak the full CIFS ?

Or is application using old API's so that it cannot cope with with long file names ?

I'm sorry that we don't have the answers you're looking for. The reason is that the Celerra is not making the change to the filenames. Your application is changing the name at some point. I suspect that Rainer is correct in that it may be an older API which was coded for ASCII filesystems and is not communicating correctly with UTF-8 filesystems.

Please consult with your application vendor to locate the source of the changing filenames.

Actually we do do some name mangling in certain cases

one being that when a Windows file with a long name is created we have to also generate a DOS M83 (8.3) file name - otherwise an application using DOS style I/O would have problem.

Thats the same as any Windows box will do

Another example would be if you are creating a file from Unix NFS with characters that are illegal in a Windows CIFS file name or when we do UniCode translation.

Are you sure Unicode (I8N) is setup properly ?

I suggest you open a service request and provide more info. It should be possible - maybe with a network trace - to find out how these file names are changed.

Technically, you are correct when saying that we do some name mangling as in when we create an M83 name from a long filename, but nothing on the order of what Marvin is detailing.

Before you open a Service Request, I would suggest you run a network capture on the Celerra and run it through Wireshark. You should be able to see which host is issuing the rename command and track it down that way. If you find that there is no host issuing renames, and the name is changing on the Celerra, definitely open a Service Request.

Let me know if you need assistance with running a tcpdump on the Data Mover.