Have you tried writing to HDFS? If your namenode is up, HDFS will appear to be quite healthy until it times out your datanode and decides that it is down.
You can see these errors if HDFS is just overwhelmed, but they will be intermittent. -Eric On Fri, Mar 8, 2013 at 10:04 AM, Mike Hugo <[email protected]> wrote: > It doesn't appear to be down - I can hit http://server:50070/dfshealth.jsp- > doesn't show any errors. Also scans work correctly and the system is > returning data normally. Other than those two errors in the logs, > everything appears to be working (?). > > Am I correct in assuming that if HDFS was down, we would be seeing more > problems than just the replication error and minc error? > > Mike > > > > > On Fri, Mar 8, 2013 at 8:19 AM, Eric Newton <[email protected]> wrote: > >> Bring HDFS back up. >> >> -Eric >> >> >> On Fri, Mar 8, 2013 at 9:17 AM, Mike Hugo <[email protected]> wrote: >> >>> We have a test server that is just a single machine accumulo instance. >>> This morning it began reporting the following error: >>> >>> java.io.IOException: File /accumulo/tables/!0/table_info/F0001ac7.rf_tmp >>> could only be replicated to 0 nodes, instead of 1 at >>> >>> I also noticed one other table is failing on MinC >>> MinC failed (java.io.IOException: File >>> /accumulo/tables/28/t-0001a92/F0001ac9.rf_tmp could only be replicated to 0 >>> nodes, instead of 1 >>> >>> How do I resolve these errors? >>> >>> Thanks! >>> >>> Mike >>> >> >> >
