Thanks. I should have checked Jira first. sorry. bwolen
On 6/10/07, Nigel Daley <[EMAIL PROTECTED]> wrote:
This is a known issue: http://issues.apache.org/jira/browse/HADOOP-1212 Nige On Jun 8, 2007, at 5:35 PM, Bwolen Yang wrote: > Looks like this problem is not related to the upgrade, but a > problem in 0.13.0. > Basically if I were to delete all the dfs directories and reformat, > the datanodes come up ok. however, if I were to reformat again, the > datanodes won't be able to come up due to the exception mentioned > below. > > bwolen > > On 6/8/07, Bwolen Yang <[EMAIL PROTECTED]> wrote: >> Hi. >> >> Upgrading from 0.12.3 to 0.13 seems to work fine. (at least the >> before and after "fsck "and "dfs -lsr" outputs matches). >> >> Then I shut down the DFS cluster, and reformat the DFS (to start a >> new >> round of performance test). The datanodes all have the following >> exception. >> >> 2007-06-08 17:13:27,985 ERROR org.apache.hadoop.dfs.DataNode: >> java.io.IOExceptio >> n: Incompatible namespaceIDs in /mnt/hadoop/bwolen/dfs/data: namenode >> namespaceID = 1399119814; datanode namespaceID = -1646824738 >> >> I'll blow away these directories and start from scratch. Just >> thought >> to let you know. >> >> bwolen >>
