For the second time in less than a week I have had to restart my Windows 2000 server and the common thread seems to be 97-99% CPU usage by pervasive (ntdbmgr.exe).
Does anyone have any insight on this or suggestions on how to troubleshoot ?

Check this articles at the Pervasive Website
 
http://www.pervasive.com/csearch/results.asp?q=CPU+Usage&c_3=on&c_7=on&a mp;c_4040=on&c_6=on&c_4=on&c_5=on&c_2=on&t=-360&s=
 
Kelloggs

We have 18 licenses but rarely have more than 6 to 10 users on at a time.  As I said, our server is Windows 2000, workstations are a mix of W2000 and XP.  Pervasive version 8.5.
We have not had this happen in quite some time.  
The only thing I have changed since I got back from the symposium is to give everyone the IN-A T7 version.  On Last Thursday I saw that the memory usage was high, 183M.  This is what I have noticed in the past when I had to restart the server.  My tech support believes there is a "memory leak" problem with pervasive, and I wrote about this some time ago here in the forum and never got a response.  I am not even sure I understand that term.
I did not check the CPU usage on Thursday.  Yesterday the memory usage was around 100M, and CPU usage was 99.
We also connect through ODBC to Access reports.  I don't know if people were using those reports yesterday when we started having problems.

Quote from cathyh on 03/11/08 at 14:48:07:

 
183MB and 100MB is NOT high RAM usage. Now, 1.5GB, that would be high RAM usage. Something tells me that you don't run your Pervasive server on a machine with less than 512MB of RAM. If you do then you should be wacked in the knuckles with a ruler.  
 
A memory leak is a problem where a program claims some RAM and then forgets about it and the RAM stays claimed for as long as the program is running. This can become very bad because the program will probably claim more memory later on and also forget about that memory too. Eventually this can really add up and start to claim most all of the RAM in the machine. Rebooting resets the count to zero but if you continue to run the leaky program it will most likely continue to leak.
 
Personally, I'd suspect the ODBC to Access reports. I've seen the pervasive system lock up at 100% CPU load when trying to interface UPS Worldship with Evo. When Worldship tries to access the Evo files it will lock up the Pervasive client for quite some time.

Double check that ALL machines accessing pervasive server have the same version as the server. We had this happen when different versions were in use. It would be very crashy - locky type of attitude also.
 
John
 

There is a Pervasive SP 8.7 that I vaguely recall addressed something with memory usage.  I do recall a couple of people on 8.5 installing it and performance improved.  I would recommend it.  You can still download it from the Pervasive web site.

We had that exact issue a few years back, on Version 8.5.  The upgrade to 8.7 did fix it, so hopefully it will do the same for you.  See  
 
http://www.istechforum.com/YaBB.pl?num=1114454988/0
 
for more on the topic.