I have ran the full set of unit tests on the patched version and I see 3
failures in HDFS HA stress tests, which are clearly not related to fix in
question. I will commit it shortly.
On Thu, May 16, 2013 at 12:03AM, Roman Shaposhnik wrote:
> On Wed, May 15, 2013 at 11:09 PM, Vinod Kumar Vavilapalli
> <[EMAIL PROTECTED]> wrote:
> > It's a little dirty, but Mark and Jarek (presumably from BigTop) ran a patched version with my change at MAPREDUCE-5240.
> I've just updated the patch to apply cleanly on branch-2.0.4 (had to refactor
> quite a bit of unit tests). Hopefully Cos can apply it tomorrow and we
> can start testing Bigtop.
> > Otherwise we may end up creating lots of bug-fix releases.
> Agreed. The point of 188.8.131.52 is to unblock Bigtop 0.6.0 release. We're
> trying to come with stable base line for Hadoop 2.0.x where
> all the downstream components will be running smoothly.
> At this point a MAPREDUCE-5240 is the only thing holding us
> up, but once we start our own round of testing other issues
> could pop up. We shall see how it goes.