Order by clause with Hive causes performance degradation.

[grafke]

Order by clause with Hive causes performance degradation. It doesn't really do anything else except on some cases causes queries to fail. I'm not sure if this is the right way to fix it, but on the other hand disabling order by really helps with performance. Otherwise, perhaps it would be wise to add an order by button to the web interface. So that the user could control it.

[landscape-bot]

Repository health decreased by 0.14% when pulling e791bff on grafke:master into 2f64c42 on airbnb:master.

• 1 new problem was found (including 0 errors and 1 code smell).
• No problems were fixed.

[coveralls]

Coverage decreased (-0.004%) to 81.58% when pulling 713d895 on grafke:master into 2f64c42 on airbnb:master.

[mistercrunch]

Wait. That order_by is there for a reason, it enforces the "Series limit". Also using Hive as a Caravel backend is probably not a great experience. You probably want to use Presto or Impala instead.

[coveralls]

Coverage decreased (-0.004%) to 81.58% when pulling 713d895 on grafke:master into 2f64c42 on airbnb:master.

[landscape-bot]

Code quality remained the same when pulling 713d895 on grafke:master into 2f64c42 on airbnb:master.

[grafke]

I see, yes, good point. I agree Hive is not the best choice. Would it be OK to add an if clause, i.e. if Hive then don't do the ordering?

[mistercrunch]

Series Limit would then act as randomly selected set of series, and only for Hive? Makes no sense to me.

[grafke]

You're absulutely right, but the subselect on Hive at the moment does not
work anyway. Any plans on adding Hive support? I'm sure we could find a way
to make it work with Hive.
On Thu, 12 May 2016 at 02:50, Maxime Beauchemin notifications@github.com
wrote:

Series Limit would then act as randomly selected set of series, and only
for Hive? Makes no sense to me.

—
You are receiving this because you authored the thread.
Reply to this email directly or view it on GitHub
#454 (comment)