[
https://issues.apache.org/jira/browse/CASSANDRA-1119?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12877474#action_12877474
]
Jonathan Ellis commented on CASSANDRA-1119:
-------------------------------------------
Is it possible that we could end up with garbage in the commitlog header, if we
lose power during the header update post flush? (i.e., not a brand new
segment?)
if so we should probably split the header out into a separate file, and if we
can't read the header, start replay of the mutations file from the beginning.
(This could be a new ticket.)
> detect incomplete commitlogheader
> ---------------------------------
>
> Key: CASSANDRA-1119
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1119
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Reporter: Jonathan Ellis
> Assignee: Matthew F. Dennis
> Priority: Minor
> Fix For: 0.6.3
>
> Attachments: 0001-cassandra-0.6-1119.patch, 0001-trunk-1119.patch,
> 0002-cassandra-0.6-1119.patch, 0002-trunk-1119.patch
>
>
> Kelvin reported:
> I just came across a corrupted CL file. Here's the stacktrace when starting
> the server:
> Listening for transport dt_socket at address: 8888
> java.lang.reflect.InvocationTargetException
> at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> at
> sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:39)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
> at java.lang.reflect.Method.invoke(Method.java:597)
> at
> org.apache.commons.daemon.support.DaemonLoader.load(DaemonLoader.java:160)
> Caused by: java.io.EOFException
> at java.io.RandomAccessFile.readInt(RandomAccessFile.java:725)
> at java.io.RandomAccessFile.readLong(RandomAccessFile.java:758)
> at
> org.apache.cassandra.db.commitlog.CommitLogHeader.readCommitLogHeader(CommitLogHeader.java:145)
> at
> org.apache.cassandra.db.commitlog.CommitLog.recover(CommitLog.java:181)
> at
> org.apache.cassandra.db.commitlog.CommitLog.recover(CommitLog.java:167)
> at
> org.apache.cassandra.thrift.CassandraDaemon.setup(CassandraDaemon.java:95)
> at
> org.apache.cassandra.thrift.CassandraDaemon.init(CassandraDaemon.java:142)
> ... 5 more
> He added that the segment is only 6 bytes long, indicating that the header
> was never completely written. CLH should catch that EOF and skip the segment
> when replay is attempted.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
