Hi,
> A rather long import process is stuck since ~12hrs on a query that
> normally returns in a few seconds or minutes. Kust like to OP there's
> almost no activity on the box, only a single postgres process at 100% CPU.
>
> Two separate queries, (although similar) have triggered the issue, and
> they are run one after another. Also, the import is also not always
> failing, I think it happened twice in about 10 runs. We're expected to
> go live in about two weeks so these were only test runs.
>
> The explain looks like this:
> http://explain.depesz.com/s/TqD
The query eventually completed in more than 18h. For comparison a normal
run doesn't take more than 1m for that specific step.
Do you think that bad stats and suboptimal plan alone could explain such
a behaviour?
Cheers
--
Matteo Beccati
Development & Consulting - http://www.beccati.com/