We have a Pervasive database running AccPac. I use RBase with ODBC connections connecting to the Pervasive database to a great extent. Both to retrieve and update data in the Pervasive database.
This has been working for 3 years without any issues and running quite well.
Two weekends ago, I did a dump and reload of the AccPac data. I also upgraded the version of Rbase that is being used. (I would not have thought this would make a difference, since it is using the same Pervasive ODBC driver)
Now the file server running ntdbsmgr.exe starts eating memory and eventually crashes.
Using PERFMON and virtual bytes monitor on ntdbsmgr.exe, all starts out OK. I will then
see occasional spikes on the monitor and it never releases. If I stop and restart Pervasive services, the memory resets to zero.
The unfortunate part is I cannot track what is causing the spikes but more importantly why Pervasive is not releasing the memory. I have went a day without issue, but more often I have to stop and restart Pervasive 2-3 times per day. This never happened before the dump/reload/Rbase upgrade. (No programs changes in the Rbase, just a database engine upgrade)
We have less than 12 users accessing the Pervasive via AccPac at any given time.
We have up to 45 users connected via ODBC to the Pervasive database.
Pervasive version 9.71 with compatibility 9.0 checked.
Transaction durability turned on
System cache turned on
Allocate at startup not turned on
Back to minimal with no activity turned on
Microkernel turned off on all machines other than the file server.
File server does not run any other software except the Pervasive server engine.
File server has 4 gig memory.
Once the Perf. Monitor shows ntdbsmgr,exe hitting 2 Gig, it crashes, giving a program exception.