Can you check namenode log around the time 'Failed to close inode' error was thrown ?
Thanks On Mon, Dec 1, 2014 at 4:10 AM, Robert Kent <[email protected]> wrote: > > Looks like an HDFS issue. Are you sure your HDFS is working fine? > > HDFS appears to be working correctly - HBase will process requests > properly and everything appears to work correctly for hours/days, until the > regionserver randomly falls over. If there were HDFS issues I would expect > to see these during normal operation, but I don't. >
