I forgot to mention about importance point: it's a blocker issue to compile Hadoop with JDK8. Hence, we need to merge it on both client side and server slide anyway.
Thanks, - Tsuyoshi On Wed, Jun 29, 2016 at 12:24 PM, Tsuyoshi Ozawa <[email protected]> wrote: > Thanks Vinod, Sangjin, Sean for your comment. > > Okay, I will take a look at the class path isolation. > Should I postpone to merge Guice upgrade to trunk? IMHO, it works with > tests, so it's okay to merge to runk. Thoughts? > > - Tsuyoshi > > On Wed, Jun 29, 2016 at 12:10 PM, Sangjin Lee <[email protected]> wrote: >> Yeah it would be awesome if we can get feedback and/or suggestions on these >> JIRAs (HADOOP-11804 and HADOOP-13070). >> >> Thanks, >> Sangjin >> >> On Wed, Jun 29, 2016 at 10:55 AM, Sean Busbey <[email protected]> wrote: >>> >>> At the very least, I'm running through an updated shaded hadoop client >>> this week[1] (HBase is my test application and it wandered onto some >>> private things that broke in branch-2). And Sangjin has a good lead on >>> an lower-short-term-cost incremental improvement for runtime isolation >>> of apps built on yarn/mapreduce[2]. He's been patiently waiting for >>> more review feedback. >>> >>> >>> [1]: https://issues.apache.org/jira/browse/HADOOP-11804 >>> [2]: https://issues.apache.org/jira/browse/HADOOP-13070 >>> >>> On Wed, Jun 29, 2016 at 12:33 PM, Vinod Kumar Vavilapalli >>> <[email protected]> wrote: >>> > My strong expectation is that we’ll have a version of classpath >>> > isolation in our first release of 3.x. I’m planning to spending some >>> > cycles >>> > right away on this. >>> > >>> > Assuming classpath isolation gets in, it is reasonable to bump up our >>> > dependencies like Jetty / Guice to the latest stable versions. >>> > >>> > Thanks >>> > +Vinod >>> > >>> >> On Jun 27, 2016, at 6:01 AM, Tsuyoshi Ozawa <[email protected]> wrote: >>> >> >>> >> Hi developers, >>> >> >>> >> I will plan to upgrade Google Guice dependency on trunk. The change >>> >> also includes asm and cglib upgrade. >>> >> I checked following points: >>> >> >>> >> * Both HDFS and YARN UIs work well. >>> >> * All webIU-related tests pass as described on HADOOP-12064. >>> >> * Ran mapreduce job, and it works well. >>> >> >>> >> https://issues.apache.org/jira/browse/HADOOP-12064 >>> >> >>> >> Do you have any concern or opinion? I would like to merge it to trunk >>> >> on this Friday if you have no objections. >>> >> >>> >> Best, >>> >> - Tsuyoshi >>> >> >>> >> --------------------------------------------------------------------- >>> >> To unsubscribe, e-mail: [email protected] >>> >> For additional commands, e-mail: [email protected] >>> >> >>> >> >>> > >>> > >>> > --------------------------------------------------------------------- >>> > To unsubscribe, e-mail: [email protected] >>> > For additional commands, e-mail: [email protected] >>> > >>> >>> >>> >>> -- >>> busbey >>> >>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: [email protected] >>> For additional commands, e-mail: [email protected] >>> >> --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
