Re: IPv6 Feature branch
Dhruba's off in a different timezone. So he might not be able to. I'll ask though. Nate, I, and Nemanja will be working on this. We're really hoping the patch will be living for just a few weeks or a month at most. Hopefully a good deal of the code we write will be testable so that will result in some good clean up. Though it won't be a primary top line goal. On Wed, Aug 19, 2015 at 6:34 PM, Steve Loughran wrote: > > +1 for a branch; ideally not too long lived. Dhruba still has commit > rights, perhaps he could be persuaded to help with it. > > If you look at the Hadoop code, it's not just network assumptions, it's > fairly brittle to bad network setup -and not helpful when these situations > arise. What could be good as part of this/a slideline is to have some entry > point where you can probe network setup, failing fast (with an error code) > if a condition isn't met (e.g. the JVM doesn't get an IPv6 address, or its > address doesn't match that you get when you look up the hostname) > > > On 17 Aug 2015, at 17:04, Elliott Clark wrote: > > > > Nate (nkedel) and I have been working on IPv6 on Hadoop and HBase lately. > > We're getting somewhere but there are a lot of different places that make > > assumptions about network. That means there will be a good deal of follow > > on patches as we find more and more places that need some TLC. > > > > Would a feature branch be good here so that we can move quickly without > > hurting stability until all of the issues are done? > > > > Thoughts? Comments? > > > > Thanks > >
Re: IPv6 Feature branch
+1 for a branch; ideally not too long lived. Dhruba still has commit rights, perhaps he could be persuaded to help with it. If you look at the Hadoop code, it's not just network assumptions, it's fairly brittle to bad network setup -and not helpful when these situations arise. What could be good as part of this/a slideline is to have some entry point where you can probe network setup, failing fast (with an error code) if a condition isn't met (e.g. the JVM doesn't get an IPv6 address, or its address doesn't match that you get when you look up the hostname) > On 17 Aug 2015, at 17:04, Elliott Clark wrote: > > Nate (nkedel) and I have been working on IPv6 on Hadoop and HBase lately. > We're getting somewhere but there are a lot of different places that make > assumptions about network. That means there will be a good deal of follow > on patches as we find more and more places that need some TLC. > > Would a feature branch be good here so that we can move quickly without > hurting stability until all of the issues are done? > > Thoughts? Comments? > > Thanks
Re: IPv6 Feature branch
+1 I don't suppose you could be conned into fixing multi-NIC and other networking issues also? ;) Do you have a list of contributors who plan to work on this feature? -C On Mon, Aug 17, 2015 at 5:04 PM, Elliott Clark wrote: > Nate (nkedel) and I have been working on IPv6 on Hadoop and HBase lately. > We're getting somewhere but there are a lot of different places that make > assumptions about network. That means there will be a good deal of follow > on patches as we find more and more places that need some TLC. > > Would a feature branch be good here so that we can move quickly without > hurting stability until all of the issues are done? > > Thoughts? Comments? > > Thanks
Re: IPv6 Feature branch
+1, would be great to see Hadoop get ipv6 support. Colin On Mon, Aug 17, 2015 at 5:04 PM, Elliott Clark wrote: > Nate (nkedel) and I have been working on IPv6 on Hadoop and HBase lately. > We're getting somewhere but there are a lot of different places that make > assumptions about network. That means there will be a good deal of follow > on patches as we find more and more places that need some TLC. > > Would a feature branch be good here so that we can move quickly without > hurting stability until all of the issues are done? > > Thoughts? Comments? > > Thanks
Re: IPv6 Feature branch
Absolutely. On Mon, Aug 17, 2015 at 5:04 PM, Elliott Clark wrote: > Nate (nkedel) and I have been working on IPv6 on Hadoop and HBase lately. > We're getting somewhere but there are a lot of different places that make > assumptions about network. That means there will be a good deal of follow > on patches as we find more and more places that need some TLC. > > Would a feature branch be good here so that we can move quickly without > hurting stability until all of the issues are done? > > Thoughts? Comments? > > Thanks >