Thanks for the responses lads. Sounds good.
I took the tip of HBASE-10070 for a spin just now. It started up fine over
existing dataset. At first it threw me off because it is missing trunk
features but after I got over that, it looked fine. No strange errors in
logs using my config from pre-HBASE-10070 startup. Thats good.
Running a pure randomread test on a small cluster I seem to see a slight
slowdown (14.2k/second vs 13.7k/second out of bucketcache). You seen that?
On Tue, May 27, 2014 at 5:27 PM, Enis Söztutar <[EMAIL PROTECTED]> wrote: