On Tue, Feb 22, 2005 at 10:41:48AM -0600, Jeff Schoby wrote: > > Extracted, compiled, and installed form samba-latest tarball source. > > We have an application running on a windows 2000 server that mapps a > drive to HEATHER. This application will periodically poll the mapped > drive for files (*.sql - small text files with an MSSQL query in them), > read them, execute the query to an MSSQL server and then delete the > file. This directory can have anywhere from 0-30,000 files in it at any > given time. Whenever the process on the windows 2000 server is > accessing the samba drive the smbd process servicing it pretty much eats > up the cpu on the samba server. The process on the win2k server goes > through the files on the samba drive very slowly no matter how many > files are in it. > > Any ideas/suggestions?
Ok - for a directory with 30,000 files in it you need to have the latest svn source for Samba 3.0 (this fix isn't in 3.0.11 I'm afraid as it was deemed too big a change at too late a stage) and to follow the advice I gave on the list about how to set up a share to efficiently cope with large directories. Someone from IBM is happily using this for < 100,000 file directories, so I'm comfortable that it works. Jeremy. -- To unsubscribe from this list go to the following URL and read the instructions: https://lists.samba.org/mailman/listinfo/samba