Oh let me take a look at this interesting issue, haven't noticed it yet... But ideally, I think HBASE-20952 is the correct way to decouple with the hadoop filesystem, that we first decouple WAL system with the hadoop filesystem, then we can abstract a 'FileSystem' for storing HFile, which could be very simple, as in HBase we only rely on a very small set of file system APIs for storing HFile, and then maybe we can just implement it based on the S3 API directly...
And +1 on landing it to hbase-filesystem repo first. Zach York <[email protected]> 于2019年4月27日周六 上午6:05写道： > +1 > > On Fri, Apr 26, 2019 at 2:48 PM Andrew Purtell <[email protected]> > wrote: > > > +1 > > > > On Fri, Apr 26, 2019 at 7:33 AM Sean Busbey <[email protected]> wrote: > > > > > Hi folks! > > > > > > HBASE-22149 "HBOSS: A FileSystem implementation to provide HBase's > > > required semantics" has an approach to reliably running HBase on top > > > of non-HDFS Filesystem implementations. > > > > > > Copying from my comment there: > > > > > > > Given the combination of this a) needing hadoop 3 only and b) being > an > > > experimental approach that we're not sure on sustainability in > production > > > I'd much prefer a different repository. > > > > > > > > Is anyone opposed to landing this in a new repository, i.e. > > > `hbase-filesystem`? Provided it includes instructions for installation > / > > > set up we wouldn't even need to add the artifacts from that repository > > as a > > > dependency for the main repo's binary artifacts. > > > > > > > > > -- > > Best regards, > > Andrew > > > > Words like orphans lost among the crosstalk, meaning torn from truth's > > decrepit hands > > - A23, Crosstalk > > >
