Assuming your database was not corrupted in the first place and the error is reproducable, it would help to see the stack trace. TO get it on a Linux box, use 'ps' to get the process id of the database server, call 'gdb mserver <processid>' and let it continue to run. Then re-issue the query, which should lead to gdb to report on the error. The command 'thr apply all where' would show what all threads are doing during the crash. thanks, Martin On 1/27/13 8:37 PM, Brian Hood wrote:
Yes i'm always joining on the guid field which is the primary key on all 3 tables.
Ok thankyou initial findings are as follows tablec is by no means correct which i'll have to investigate however it still shouldn't of crashed the database but maybe of causing some serious main memory allocation trying to join it.
however tableb also isn't quite correct either about 300 or so rows out.