[ 
https://issues.apache.org/jira/browse/CASSANDRA-7664?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Yuki Morishita updated CASSANDRA-7664:
--------------------------------------

    Reviewer: Yuki Morishita

> IndexOutOfBoundsException thrown during repair
> ----------------------------------------------
>
>                 Key: CASSANDRA-7664
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-7664
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>         Environment: RHEL 6.1
> Casandra 1.2.3 - 1.2.18
>            Reporter: ZhongYu
>            Assignee: Marcus Eriksson
>             Fix For: 2.0.10
>
>         Attachments: $O[TOX~GGUZRW~IHPYPEG{0.jpg, 
> 0001-throw-EOFException-if-we-run-out-of-chunks-in-compre.patch, 
> other-errors.txt
>
>
> I was running repair command with moderate read and write load at the same 
> time. And I found tens of IndexOutOfBoundsException in system log as follows:
> {quote}
> ERROR [Thread-6056] 2013-05-22 14:47:59,416 CassandraDaemon.java (line132) 
> Exception in thread Thread[Thread-6056,5,main]
> java.lang.IndexOutOfBoundsException
>         at sun.nio.ch.ChannelInputStream.read(ChannelInputStream.java:75)
>         at 
> org.apache.cassandra.streaming.compress.CompressedInputStream$Reader.runMayThrow(CompressedInputStream.java:151)
>         at 
> org.apache.cassandra.utils.WrappedRunnable.run(WrappedRunnable.java:28)
>         at java.lang.Thread.run(Thread.java:662)
> {quote}
> I read the source code of CompressedInputStream.java and found there surely 
> will throw IndexOutOfBoundsException in the following situation:
> {code:title=CompressedInputStream.java|borderStyle=solid}
> // Part of CompressedInputStream.java start from Line 139
>         protected void runMayThrow() throws Exception
>         {
>             byte[] compressedWithCRC;
>             while (chunks.hasNext())
>             {
>                 CompressionMetadata.Chunk chunk = chunks.next();
>                 int readLength = chunk.length + 4; // read with CRC
>                 compressedWithCRC = new byte[readLength];
>                 int bufferRead = 0;
>                 while (bufferRead < readLength)
>                     bufferRead += source.read(compressedWithCRC, bufferRead, 
> readLength - bufferRead);
>                 dataBuffer.put(compressedWithCRC);
>             }
>         }
> {code}
> If read function read nothing because the end of the stream has been reached, 
> it will return -1, thus bufferRead can be negetive. In the next circle, read 
> function will throw IndexOutOfBoundsException because bufferRead is negetive. 



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to