Some quick checks: - Are configuration files consistent across your DNs (w.r.t. NameNode location [fs.default.name], primarily)? - Do you have a firewall running that may be blocking out connections from a slave to the master over the specified IPC ports?
On Thu, Apr 28, 2011 at 10:14 PM, Fabio Souto <[email protected]> wrote: > Hello, > > I'm having some problems setting up my datanodes, I have a 4 node cluster > (all of them are datanodes), if I run > > sudo -u hdfs hadoop dfsadmin -report > > > Configured Capacity: 112231907328 (104.52 GB) > Present Capacity: 54121451520 (50.4 GB) > DFS Remaining: 54120173568 (50.4 GB) > DFS Used: 1277952 (1.22 MB) > DFS Used%: 0% > Under replicated blocks: 1 > Blocks with corrupt replicas: 0 > Missing blocks: 0 > > ------------------------------------------------- > Datanodes available: 1 (1 total, 0 dead) > > Name: <my-ip>:50010 > Decommission Status : Normal > Configured Capacity: 112231907328 (104.52 GB) > DFS Used: 1277952 (1.22 MB) > Non DFS Used: 58110455808 (54.12 GB) > DFS Remaining: 54120173568(50.4 GB) > DFS Used%: 0% > DFS Remaining%: 48.22% > Last contact: Thu Apr 28 18:39:28 CEST 2011 > > > The report only show 1 datanode! Checking the logs of the slaves I found this: > > > 2011-04-28 18:26:09,587 INFO org.apache.hadoop.ipc.Client: Retrying connect > to server: slave/<ip>:54310. Already tried 9 time(s). > 2011-04-28 18:26:09,588 INFO org.apache.hadoop.ipc.RPC: Server at > slave/<ip>:54310 not available yet, Zzzzz... > > > I don't know what to do....Should I configure passwordless ssh between the > servers? > > Thanks -- Harsh J
