[OpenAFS-devel] Apple Xserve ....

[Hartmut Reuter]

You should use the NAMEI-Interface instead of the old inode-Interface 
which is difficult and dangerous to implement on new platforms. Make 
sure the param.h contains a define for AFS_NAMEI_ENV. With this 
interface the fileserver doesn't need any kernel extensions.

Hartmut

Valerio Luccio wrote:
> Still having lots of problems getting the AFS server to work on an Apple 
> Xserve.
> 
> More often than not, when I modify a file created on a volume (for 
> example if I try to delete it) mounted under root.cell, the volume and 
> the root.cell volume get corrupted (I get an "Cannot attach volume" when 
> I do an "vos listvol"). I've restarted the fileserver process with the 
> maximum debugs turned on and I notice the following lines in the FileLog:
> 
>    SAFS_RemoveFile <filename>, Did=<number>, Host <IP address>, Id 1
>    DT: inode=202, name=<filename>, errno=17
> 
> I tracked the "DT:" line to the DeleteTarget function in 
> "afsfileprocs.c" right after the calls to IH_DEC and IH_RELEASE. The 
> error seems to be set by by IH_DEC.
> 
> If I read the source correctly, for Darwin this corresponds to a call to 
> "idec" (in afssyscalls.c"), which is actually a call to "syscall" with 
> the first parameter set to "71+1000". Am I correct so far ? If I am, 
> then I'm lost at this point. The only "syscall" source code I see is 
> assembler code which would seem to be defined only for AIX, SGI and for 
> something defined as "__hp9000s800" (whatever that is).
> 
> Does anyone have any idea ?
> 
> Thanks.
> 


-- 
-----------------------------------------------------------------
Hartmut Reuter                           e-mail reuter@rzg.mpg.de
					   phone +49-89-3299-1328
RZG (Rechenzentrum Garching)               fax   +49-89-3299-1301
Computing Center of the Max-Planck-Gesellschaft (MPG) and the
Institut fuer Plasmaphysik (IPP)
-----------------------------------------------------------------