[HDFS-14615] NPE writing edit logs

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 3.1.1
• Component/s: namenode
• Assignee: Lisheng Sun
• Reporter: Wei-Chiu Chuang
• Created: 27/Jun/19 16:20
• Updated: 02/Oct/19 17:12

Description

Hit a weird bug where writing mkdir op to edit log throws an NPE and NameNode crashed

2019-06-26 10:57:27,398 ERROR org.apache.hadoop.hdfs.server.namenode.FSEditLog: Error: write op failed for (journal JournalAndStream(mgr=FileJournalManager(root=/ssd/work/src/upstream/impala/testdata/cluster/cdh6/node-1/data/dfs/nn), stream=EditLogFileOutputStream(/ssd/work/src/upstream/impala/testdata/cluster/cdh6/node-1/data/dfs/nn/current/edits_inprogress_0000000000000598588)))
java.lang.NullPointerException
        at org.apache.hadoop.io.Text.encode(Text.java:451)
        at org.apache.hadoop.io.Text.encode(Text.java:431)
        at org.apache.hadoop.io.Text.writeString(Text.java:491)
        at org.apache.hadoop.fs.permission.PermissionStatus.write(PermissionStatus.java:104)
        at org.apache.hadoop.fs.permission.PermissionStatus.write(PermissionStatus.java:84)
        at org.apache.hadoop.hdfs.server.namenode.FSEditLogOp$MkdirOp.writeFields(FSEditLogOp.java:1654)
        at org.apache.hadoop.hdfs.server.namenode.FSEditLogOp$Writer.writeOp(FSEditLogOp.java:4866)
        at org.apache.hadoop.hdfs.server.namenode.EditsDoubleBuffer$TxnBuffer.writeOp(EditsDoubleBuffer.java:157)
        at org.apache.hadoop.hdfs.server.namenode.EditsDoubleBuffer.writeOp(EditsDoubleBuffer.java:60)
        at org.apache.hadoop.hdfs.server.namenode.EditLogFileOutputStream.write(EditLogFileOutputStream.java:97)
        at org.apache.hadoop.hdfs.server.namenode.JournalSet$JournalSetOutputStream$1.apply(JournalSet.java:444)
        at org.apache.hadoop.hdfs.server.namenode.JournalSet.mapJournalsAndReportErrors(JournalSet.java:385)
        at org.apache.hadoop.hdfs.server.namenode.JournalSet.access$100(JournalSet.java:55)
        at org.apache.hadoop.hdfs.server.namenode.JournalSet$JournalSetOutputStream.write(JournalSet.java:440)
        at org.apache.hadoop.hdfs.server.namenode.FSEditLog.doEditTransaction(FSEditLog.java:481)
        at org.apache.hadoop.hdfs.server.namenode.FSEditLogAsync$Edit.logEdit(FSEditLogAsync.java:288)
        at org.apache.hadoop.hdfs.server.namenode.FSEditLogAsync.run(FSEditLogAsync.java:232)

The stacktrace is similar to SENTRY-555, which is thought to be a Sentry bug (authorization provider), but this cluster doesn't have Sentry and therefore could be a genuine HDFS bug.

File this jira to keep a record.