+1 Thanks Konstantin for driving the merge vote!
I have been working on the development and testing of the feature. It has been running for several weeks on our cluster with ~100 nodes, which has HA and Kerberos enabled. I have been able to run several different MapReduce jobs and HDFS benchmarks (see HDFS-14058 for more detail). I feel confident that the feature now has the completed functionality and is ready for merge into trunk. Chen Konstantin Shvachko <shv.had...@gmail.com> 于2018年12月14日周五 下午6:24写道: > Hi Hadoop developers, > > I would like to propose to merge to trunk the feature branch HDFS-12943 for > Consistent Reads from Standby Node. The feature is intended to scale read > RPC workloads. On large clusters reads comprise 95% of all RPCs to the > NameNode. We should be able to accommodate higher overall RPC workloads (up > to 4x by some estimates) by adding multiple ObserverNodes. > > The main functionality has been implemented see sub-tasks of HDFS-12943. > We followed up with the test plan. Testing was done on two independent > clusters (see HDFS-14058 and HDFS-14059) with security enabled. > We ran standard HDFS commands, MR jobs, admin commands including manual > failover. > We know of one cluster running this feature in production. > > Since the previous vote we addressed Daryn's concern (see HDFS-13873), > added documentation for the new feature, and fixed a few other jiras. > > I attached a unified patch to the umbrella jira for the review. > Please vote on this thread. The vote will run for 7 days until Wed Dec 21. > > Thanks, > --Konstantin >