Josh Elser created ACCUMULO-1846:
------------------------------------
Summary: Spurious ClassNotFoundException
Key: ACCUMULO-1846
URL: https://issues.apache.org/jira/browse/ACCUMULO-1846
Project: Accumulo
Issue Type: Bug
Reporter: Josh Elser
Fix For: 1.6.0
>From the monitor:
{noformat}
Failed to load class
java.lang.ClassNotFoundException:
org.apache.accumulo.master.recovery.HadoopLogCloser
at
org.apache.commons.vfs2.impl.VFSClassLoader.findClass(VFSClassLoader.java:175)
at java.lang.ClassLoader.loadClass(ClassLoader.java:424)
at java.lang.ClassLoader.loadClass(ClassLoader.java:357)
at
org.apache.accumulo.start.classloader.vfs.AccumuloVFSClassLoader.loadClass(AccumuloVFSClassLoader.java:102)
at
org.apache.accumulo.core.conf.AccumuloConfiguration.instantiateClassProperty(AccumuloConfiguration.java:193)
at
org.apache.accumulo.master.recovery.RecoveryManager.recoverLogs(RecoveryManager.java:162)
at
org.apache.accumulo.master.TabletGroupWatcher.run(TabletGroupWatcher.java:195)
{noformat}
>From the master.debug.log:
{noformat}
2013-11-01 23:07:26,557 [master.EventCoordinator] INFO : tablet !!R<< was
loaded on localhost:9997
2013-11-01 23:07:26,567 [master.Master] DEBUG: Finished gathering information
from 1 servers in 0.01 seconds
2013-11-01 23:07:26,568 [master.Master] DEBUG: not balancing because there are
unhosted tablets
2013-11-01 23:07:26,636 [recovery.RecoveryManager] DEBUG: Recovering
hdfs://localhost:8020/accumulo1.6/wal/localhost+9997/224c63ac-a7db-4ba6-be87-a6d1fd7a0921
to
hdfs://localhost:8020/accumulo1.6/recovery/224c63ac-a7db-4ba6-be87-a6d1fd7a0921d
to load class
at
org.apache.commons.vfs2.impl.VFSClassLoader.findClass(VFSClassLoader.java:175)
at java.lang.ClassLoader.loadClass(ClassLoader.java:424)
at java.lang.ClassLoader.loadClass(ClassLoader.java:357)
at
org.apache.accumulo.start.classloader.vfs.AccumuloVFSClassLoader.loadClass(AccumuloVFSClassLoader.java:102)
at
org.apache.accumulo.core.conf.AccumuloConfiguration.instantiateClassProperty(AccumuloConfiguration.java:193)
at
org.apache.accumulo.master.recovery.RecoveryManager.recoverLogs(RecoveryManager.java:162)
at
org.apache.accumulo.master.TabletGroupWatcher.run(TabletGroupWatcher.java:195)
2013-11-01 23:07:26,645 [conf.AccumuloConfiguration] INFO : Using
org.apache.accumulo.server.master.recovery.HadoopLogCloser
2013-11-01 23:07:26,645 [recovery.RecoveryManager] INFO : Starting recovery of
hdfs://localhost:8020/accumulo1.6/wal/localhost+9997/224c63ac-a7db-4ba6-be87-a6d1fd7a0921
(in : 10s) created for hdfs:, tablet !0;~< holds a reference
2013-11-01 23:07:26,647 [master.Master] DEBUG: 1 assigned to dead servers:
[!0<;~@(null,localhost:9997[142010edaa87210],null)]...
{noformat}
I think I was seeing this when I intentionally set the tserver's heap very
small to force it to OOM during log recovery (to verify that WAL entries aren't
prematurely deleted).
--
This message was sent by Atlassian JIRA
(v6.1#6144)
