Csaba Kalman <csaba818@freemail.hu> writes:
> Tom Lane writes:
>> Hm, what can you say about the data that goes into the three columns
>> that have hash indexes? I'm going to try testing this with random data,
>> but it'd likely be more useful if I could reproduce your data
>> distribution.
> [ some info ]
Thanks, but I've been unable to reproduce the problem with either random
data or data tailored to match yours. I'm a bit worried that the issue
may only appear on Windows --- hash indexes currently assume that the
filesystem will react in a particular way to a write that's well beyond
the current EOF, and I wouldn't put it past Microsoft to get that wrong.
Can you put together a self-contained script that causes the error for
you? It's okay if it only fails once every ten or even 100 tries.
What we need is something that other people can run and say fairly
definitively "it happens here" or "it doesn't".
regards, tom lane