Yesterday's notes: https://docs.google.com/document/d/1jXM5Ujvf-zhcyw_5kiQVx6g-HeKe-YGnFS_1-qFXomI/edit#heading=h.sgy1m4sxusyv 10/23/2019
Attendee: Weichiu, Fei Hui, Didi: a few clusters are upgraded to Hadoop 3.2. The largest production cluster will be the next. Testing Erasure Coding with decommissioning. Filed a few jiras already. Testing consistent read from standby NN, but found it requires turning off access time otherwise every block read operation becomes write operation and impact performance. However, the cluster requires access time to perform data governance. There are internal teams that rely on the access time stats. We discussed this and mentioned Uber stated the same issue in the January Hadoop meetup. Didi’s Hadoop team is interested in Ozone after all clusters are upgraded to Hadoop 3. On Wed, Oct 23, 2019 at 3:55 PM Wei-Chiu Chuang <weic...@cloudera.com> wrote: > PDT 10pm Wednesday = tonight, CST 1pm Thursday = today. > Feel free to join Zoom and chat. > > Join Zoom Meeting > > https://cloudera.zoom.us/j/880548968 > > Past sessions: > > https://docs.google.com/document/d/1jXM5Ujvf-zhcyw_5kiQVx6g-HeKe-YGnFS_1-qFXomI/edit > > Also heads-up, > On November 20/21, Feilong from Intel will present his work on HDFS-13762 > (Support non-volatile storage class memory(SCM) in HDFS cache directives). > This is going to happen in the APAC storage community sync. > > Best, > Weichiu >
