Minwoo -- any chance you could share some details about that Hadoop issue? Something on JIRA?
On Tue, Aug 15, 2017 at 8:11 PM, Kang Minwoo <[email protected]> wrote: > My team member tries to resolve this issue. > And He found it is a Hadoop Issue. > He tries to test purging FD. > > Best regards, > Minwoo Kang > > ________________________________________ > 보낸 사람: Kang Minwoo <[email protected]> > 보낸 날짜: 2017년 8월 11일 금요일 오전 11:18:58 > 받는 사람: [email protected] > 제목: Region Server does not close FD. > > Hello, HBase Users. > > These days, My team upgrade HBase Version up to 1.2.6 and test in our product. > In the meantime, My team improves real machine Fault handling. > > My server is JBOD. So My team decided using data node volume failures > thresholds in HDFS Feature. > My team set an option that is data node volume failures thresholds. > After that, My team received a call our infra team. > Because they could not unmount disk because FD was still alive. > > So My team tracked FD, and My team found Region Server had that FD. > I do not know why does Region Server not closed FD. > (My team use shortcircuit.) > > Have you ever experienced the same problem? > > My team member created a Jira Issue[1]. > > Best regards, > Minwoo Kang > > [1]: https://issues.apache.org/jira/browse/HBASE-18454
