[ https://issues.apache.org/jira/browse/MAPREDUCE-1556?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12848892#action_12848892 ]
Chris Douglas commented on MAPREDUCE-1556: ------------------------------------------ bq. It shouldn't be, should it? Avro 1.2 used to pull it in as a required dependency, which was a bug, and Common explicitly had to block it. Avro 1.3 only requires the slf4j API, not any of its implementations. Hadoop uses the log4j implementation of slf4j, which both common and mapreduce's ivy.xml explicitly require, but perhaps not in all the right contexts? I think Avro marking it as a dependency masked MR/HDFS/Common's dependence on it, which isn't explicit? It's printed whenever I try to start a MR \*Tracker from a binary tarball. > upgrade to Avro 1.3.0 > --------------------- > > Key: MAPREDUCE-1556 > URL: https://issues.apache.org/jira/browse/MAPREDUCE-1556 > Project: Hadoop Map/Reduce > Issue Type: Improvement > Components: jobtracker > Reporter: Doug Cutting > Assignee: Doug Cutting > Fix For: 0.22.0 > > Attachments: MAPREDUCE-1556.patch, MAPREDUCE-1556.patch, > MAPREDUCE-1556.patch > > > Avro 1.3.0 has now been released. HADOOP-6486 and HDFS-892 require it, and > the version of Avro used by MapReduce should be synchronized with these > projects. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.