We're running Oracle 9i release 2 (using reports 6i as the client) on a Windows 2000 advanced server (with all critical patches), the oracle databases are on a nas filer (which the windows server sees as additional disks),
however I have a problem - when I do anything with oracle .dmp files (from the import / export utility) the server crashes.
Its not very consistent - for example I took an export of our production database, imported into our test instance and it was okay, this morning I went to import it into our development instance and the server crashed!
If no errors or assciated trace files then the problem isn't within the database.
Like the event viewer says, something is pulling the carpet from under oracle, and it doesn't get a chance to raise an exception.
the oracle databases are on a nas filer (which the windows server sees as additional disks)
Thanks for your replies, I'm thinking that we may have a memory problem - the server has 4gig of ram but the instances are using a fair bit of it, the .dmp file is nearly a gig in size and I think the effort of doing anything with it is just too much.
With regards to the NAS and whether it is a supported configuration - Netapps are an Oracle partner and our application suppliers are happy with the set up.
I would agree that the NetApp filers should be fine... the thing that caguht my attention was the statement that your instances were using up almost all of the memory... you need to have a good amount of headroom, especially on a Windows machine, for connections, sort areas, server memory space, processes, etc... I would probably try and reduce the memory consumption of the instances and leave at least 1GB free to start...
1GB is not really that large of a dump file, so I wouldn't think that's the problem...
does the server blue screen? or just reboot? is there anything in the windows event logs?
you could have driver issues or just general hardware problems also, like bad memory, power supply, etc...
Thanks for the reply, the server just blue screens and reports a PCI bus to host problem.
We initially had the problem moving the databases from the server to the nas (and we did all kinds of tests on the hardware - thinking that there must be a problem), but when we moved the data in smaller chunks the problem only seemed to occur with an old .dmp file - which we assumed must have become corrupt and therefore deleted it.
Then there was the incident which led to my posting on this forum - so I started to think that windows was having a problem with Oracle .dmp files lol I was worried that somehow corrupt .dmp files were being created and causing the server to crash.
It was only after this that I started looking for alternative and more realistic solutions and realised that an awful lot of memory was being used up the windows task manager is reporting that the memory usage is 4362824 / 5598684 and the the server is dedicated to our oracle application.
We have three oracle instances running on the server and I suspect that when the application guys installed oracle etc they have allocated almost all the available memory to the Oracle instances (though i'm not sure why they didn't check with us before doing it).