If it OK to put your active namenode to safe mode, there is a way. - Put your active namenode to safe mode - Checkpoint the name space - Bootstrap Standby namenode, and start.
I think hadoop 3 allows you to use checkpoint node, it will help you to avoid this kind of issues. On Wed, Aug 28, 2019 at 5:30 AM Alexander Batyrshin <[email protected]> wrote: > Hello all, > We have an issue with free disk space that trigger > https://issues.apache.org/jira/browse/HDFS-13269 (our versions is 3.1.1). > So right now I have Active NameNode in SafeMode that can’t do checkpoint. > Second NameNode after restart is trying to build actual fsimage from > jnedits but it will take extremely long time (speed ~ 2M tx / hour and we > need to get ~ 173M). > > > Any ideas how to recover? >
