Thanks to the many responses.
My original question was
> Hello Managers,
> I have a 2100 server running 3.2b with a EISA swxcr raid controller. I
have 6 rz28's in a raid 5 configuration. I have a directory with 5000 files.
> When I do the following command on the server 'ls -l', there is a delay of
10 seconds before I start seeing files appear on the screen.
> I have also seen a problem with the display where the directory listing is
not completed and the screen stops listing the files at some arbitrary >
point.
> Are these bugs in the DU 3.2b ?
> Are there any tuning issues to be researched ?
> Or do I have to scrap this configuration and buy a PCI Raid controller
with disk and use a RAID 0+1 configuration?
I received many responses that indicated the performance problems may not be
the drive configuration but rather the command 'ls -l'. Since the 'ls -l'
creates a list first then sorts the objects and final display the files to
the screen. Some suggested that I should use 'find' to judge performance
instead of 'ls -l'. I created a simple test of "date;ls
-l>/dev/null;date;find . -name '*' -print>/dev/null;date" and found that the
'ls -l' command takes 40 seconds and the find command takes 3 seconds. I
don't believe that I have a performance problem with the disk drive
configuration. I believe it is an application problem.
Several managers have seen the same type of problem with large number of
files in a directory. The suggestion was to break down the directory into
subdirectories.
The problem I have with the screen terminating the output prematurely is a
bug. I do not know the patch number.
I do not plan to get a PCI raid controller to solve this problem. It does
not sound like it would fix this but would be nice.
Thanks for all you help and responses,
Jeff
Received on Mon Jul 01 1996 - 17:39:55 NZST