I see the same strange behavior on 2-node cluster with 0.18.3, 0.19.1 and
snv's branch-0.20.0...
2 nodes:
"master1" running NameNode, JobTracker, DataNode, TaskTracker.
"slave1" running DataNode, TaskTracker.
PROBLEM: "master" trying fetch data of "attempt" that running on slave,
BUT connecting to "localhost" for unknown reason:
(master's console:)
09/03/06 17:15:01 WARN mapred.JobClient: Error reading task
outputhttp://localhost:50060/tasklog?plaintext=true&taskid=attempt_200903061711_0001_m_000000_0&filter=stdout
But "attempt_200903061711_0001_m_000000_0" i have found in "logs/userlogs"
on "slave"!
"master" trying to fetch it, but connects to itself and, of course, can't
find if (HTTP 410)
wget
"http://localhost:50060/tasklog?plaintext=true&taskid=attempt_200903061711_0001_m_000000_0&filter=stdout"
"Failed to retrieve stderr log for task:
attempt_200903061711_0001_m_000001_0"
In the "logs/userlogs" on master there are some other "attempt"s.
(Of course, little by little all work "migrates" to "master" and all the
Job finishing successfully).