Possible TDB/Samba optimizations.

[John E. Malmberg]

COLLOT Jean-Yves wrote:
>>Dave Jones posted that he had converted the tdb routines to use indexed 
>>files for a performance improvements
> 
> Actually, it's me who did that change. Dave Jones added the automatic
> creation of indexed files when missing.

Yes, David already pointed that out to me in a private e-mail.  I was
unable to reply before this because yesterday evening my cable internet
connection was not usuable.

>>If these sharing and locking functions could be mapped to the 
>>distributed lock manager, it would work much faster
> 
> This is true, it would be far better. It was my first idea, but it's quite a
> big job to do. Anyone who has the time to do it is welcome.

I am tied up in Python right now, but I can help with looking up
information for people that want to pursue this.

I had not determined if the better approach to be would be to have
routines that used the TDB API, so that the UNIX SAMBA code would not
need to be changed, or if it would be better just to replace the UNIX
routines with OpenVMS specific routines.

Moving the sharing and locking TDB functions to the distributed lock
manager would eliminate one of the known bugs in SAMBA from the OpenVMS
platform.

If a SMBD process crashes at the wrong time, those TDB databases have
stale data, and need to be manually repaired.

-John
wb8tyw at qsl.network
Personal Opinion Only