Christian,
It looks like you are running hadoop 0.14.
This bug is related to HADOOP-1076 imo.
I recommend to shut down secondary name-node first.
And the restart the main name-node at some point when you are not at
risk to loose jobs.
You should restart the name-node with -upgrade, otherwise you can hit
HADOOP-1978,
which is now fixed in 0.14, but I don't know whether you run this patch.
Let me know if it helps.
Thanks,
--Konstantin
Christian Kunz wrote:
Looks like I ran into something like HADOOP-1978 symptom:
2007-10-04 20:51:24,197 INFO org.apache.hadoop.fs.FSNamesystem: Roll
Edit Log from <IP4>
2007-10-04 20:51:24,197 INFO org.apache.hadoop.ipc.Server: IPC Server
handler 41 on <port>, call rollEditLog() from <IP4>:35246: error:
java.io.IOException: Attempt to roll edit log but edits.new exists
java.io.IOException: Attempt to roll edit log but edits.new exists
at
org.apache.hadoop.dfs.FSEditLog.rollEditLog(FSEditLog.java:606)
at
org.apache.hadoop.dfs.FSNamesystem.rollEditLog(FSNamesystem.java:3449)
at org.apache.hadoop.dfs.NameNode.rollEditLog(NameNode.java:525)
at sun.reflect.GeneratedMethodAccessor24.invoke(Unknown Source)
at
sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessor
Impl.java:25)
at java.lang.reflect.Method.invoke(Method.java:597)
at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:340)
at org.apache.hadoop.ipc.Server$Handler.run(Server.java:609)
This goes on for 20 hours now.
Is there an easy way to salvage without losing data?
Thanks,
-Christian
