views:

35

answers:

2

Hello All,

I have been building index on a 200 million row table for almost 14 hours. Due to resource over-consumption on the machine (because of a separate incident), the machine cashed. Clearly, I want to avoid another 14 hours to re-construct the index. Is there a way that I can resume the construction of index from the point (or slightly back) where the machine crashed? I can see the temporary files created.

Thanks

A: 

I don't think there is a way to resume the index creation and quite frankly, if your machine crashed, there is no way of knowing if that temporary file contains vaild data or has been corrupted in the crash.

Gene Vincent
+1  A: 

I would guess that your priority is not getting the job done - but in having a working database with your undamaged data in it at the end. So you really need to try to repair the table first. If at the end of that you've got the index - then result. If not, then you'll need to start building the index again - and you've not really lost anything.

14 hours is rather a long time though - can you split the table and index ontp seperate disks? Is this outage sufficiently expensive to justify spending a coujple days reconfiging your hardware to get it done quicker?

C.

symcbean