Craig Mason-Jones

Hi Fred, sent you an email with a link to two example databases - did you receive them? Best, C

Hi Fred, I'm still routinely getting database corruption, although I try to routinely shutdown the database to ensure that nothing goes wrong inadvertently (power off, etc). The attached log file shows my app logs (app) interspersed with the db logs (db), and you can see at 12:05 database shutdown happens correctly, and then at 12:45 the database shutdown fails. Thereafter the database is entirely corrupted, and the application won't start again. There must be something I'm doing, but I have no idea...

Ok, I'll take it out :) But surely it wouldn't be causing the corruption?

I've got SET FILES BACKUP INCREMENT TRUE, and I SHUTDOWN COMPACT each time the application closes. It seems that systems that get corrupted are left running for a few days, though... but I can't imagine why that might trigger corruption... I also call CHECKPOINT DEFRAG each tim eI start the database. Do you think I should add a SET CHECKPOINT DEGRAG 100M, for eg?

Hi, so I worked out I was creating lots of db connections, and fixed that, and my errors have been substantially reduced, but while I had hoped they had disappeared, they are still occurring. However, I'm now logging db activity extensively. The attached log is from the hsqldb logs - you can search for the first ERROR to see where the error suddenly appears. As far as I can tell, shutdown occurred correctly before the error, or at least shutdown was initiated...

Might the IllegalStateException: zip file closed be caused by too little memory? It's still occurring on some machines, when it operates fine on my computer.

I think I've found it - a memory leak leading to memory exhaustion! (Holding thumbs, but it would explain why the db was corrupting, and why it only happens at a few sites, where I suspect they just never turn the system off...)

I'm still following up, trying to get more examples, to be sure I can send you an as-small-as-possible example. What I seem to be seeing before breakdown on most systems is this: I'm getting a java.lang.IllegalStateException: zip file closed when I call SHUTDOWN COMPACT. That seems to precede (without itself causing issues) the eventual corruption - it can be by a few days, though, during which time the db seems to work fine.