This happened to us more than once and we would like to find out why it’s doing this :
Server is running perfectly for a few hours/days and then suddenly, the Cloudscape DB seems to corrupt. .DAT file under the /packages/wmtn/data/tradingnetworks/seg0 is going from a normal few MB to a enormous 1-14 GIG, and this is only UNDER A FEW SECONDS ! It’s always this file : C1002736498544.dat .Also, there’s always a logxxx.dat file that grows to a few MBs too (instead of usual few KBs) in /tradingnetworks/log/ .When it does this, of course, the server hangs and we have to kill the java.exe process (under win2000 server).
What’s weird is that all the logs shows nothing abnormal…
- We tried to restore the DB from scratch. The same thing happened
- We reinstall IS & TN server in another directory - same thing
What can cause this to happen ?
Mathieu, do you have the option to use another DB instead of Cloudscape ? (eg. Oracle, MS SQL Server) It is not recommended to use Cloudscape in Production. There is a thread in the webMethods Advantage Trading Networks forum that includes the following response from webMethods support:
“Please note: Cloudscape is bundled, strictly, as a back-end to Trading Networks. I would also note that we do not recommend you use cloudscape in a production environment - it is a convenient option for development but does not scale well and has some stability issues that preclude it’s use in production.”
In the same thread another use mention “I have been using it in development and testing environments for a long time, and have seen it get corrupted a number of times. We had to rebuild the repository at that point, losing all existing data. You don’t want this in production.”
So switch to another rdbms if you can.
Regards,
Wayne
Yes, I have read that and aware of the “issues”. We are currently looking in using Oracle or SQL server.