[continueing]

Geoff, now about the virtual star received.

As said in the beginning, you (and the other thread) made me think from another perspective.
Let's say that it made me assume that the header would be corrupted at some point of time in my case. So I started corrupting the header ...

Let me tell you, it took a few minutes only to emulate the behaviour of the browse. Exactly the same.
Right now it is my task to find a scenario what causes the corruption I applied, which is still very very difficult because we're always facing the fact that the thing happens overnight only. But, as I explained before : once we would be able to find the cause of the browse's behaviour, we'd be near to the cause. So in fact, the overnight-thing now has some theories with it. Theories which can hardly be turned into practice, also knowing that each "test" will take another day. So main objective : make the thing turn up during the day, in order to avoid the excessive throughput time otherwise.

I know, I am not telling right now what corruption I applied in the header, which is because this specific one can't be the case. So I nead to find the "real one", and I am working on that.

For the interested, I can say a few other things about the corruption in the table itself, which I couldn't see before. This has to do with the anounced in the other message : somehow I never tried to find the beginning of the corruption by means of starting at the offset at the back : the end of the record that overflows into the the next block.

Focused on this, I easily found that the "fixed" corruption just starts at this same record in the previous block, the one being corrupted. It's almost too stupid that I didn't find this means earlier (it's over two years only). But let me tell you, it just came from the brainstorms in here (excuse me for the ever longthreaded writing ... but it just helps (me)).

[the stupid IE won't let me send this with the remainder; needs another message]