We aren't using encryption zones. The only 3.x-specific features we've enabled are storage tiering and >2 namenodes (we run 3).
________________________________ From: epa...@apache.org <epa...@apache.org> Sent: Tuesday, August 4, 2020 12:59 PM To: HDFS Dev <hdfs-dev@hadoop.apache.org>; Craig.Condit <craig.con...@target.com> Subject: [EXTERNAL] Re: 2.x client accessing 3.x HDFS and vice versa Thanks Craig. Are you enabling any special HDFS features? Specifically, are you enabling and using encryption zones? -Eric On Tuesday, August 4, 2020, 8:49:26 AM CDT, Craig. Condit <craig.con...@target.com> wrote: Eric,I can't speak to 2.10, but we have been running a production Hadoop 3.2.0 cluster for ~ 1 year now in parallel with a legacy 2.7.3 cluster, and have clients from both communicating with HDFS on the other cluster frequently.As usual, YMMV, but we haven't encountered any serious problems. - Craig Condit ________________________________ From: epa...@apache.org <epa...@apache.org> Sent: Monday, August 3, 2020 2:12 PM To: HDFS Dev <hdfs-dev@hadoop.apache.org> Subject: [EXTERNAL] 2.x client accessing 3.x HDFS and vice versa Hello, We are investigating upgrading to 3.x, but we are very concerned about the differences in the HDFS features, interfaces, etc. between 2.10 and 3.3+. Our requirements are to not have any cluster downtime and to allow 2.10 HDFS clients to communicate with 3.x clusters and 3.x HDFS clients to communicate with 2.10 clusters. Have you encountered these use cases with your users and customers and, if so, how have they addressed the issues? Thank you, -Eric Payne --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org
