[
https://issues.apache.org/jira/browse/HADOOP-15206?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16354819#comment-16354819
]
Aki Tanaka commented on HADOOP-15206:
-------------------------------------
Thank you very much for the comments!
{quote}
This doesn't just drop the first bzip2 block, it drops the entire split. This
goes back to my previous comment about the code assuming splits that start
between bytes 1-4 are always tiny. Splits do not have to be equally sized, so
theoretically there could be just two splits where the first split is a
two-byte split starting at offset 0 and the other split is the rest of the file.
{quote}
Thank you for explaining the details. I understand the problem.
{quote}The logic regarding the header seems backwards. If the header is
stripped then that means there was a header present, yet the logic is only
adding up bytes for a header length if it was not stripped which is the case
when the header is not there.
{quote}
That's right... Thank you for pointing this out.
After some tests, I noticed the following 2 points.
1. When reading from position 1-3 (on bzip2 header),
isHeaderStripped/isSubHeaderStripped is always false. This is because the
current readStreamHeader() works only when the start position is 0.
2. I set one byte beyond the start of the first bzip2 block (header_len + 1) to
the InputStream's start position, but duplicated records issue still happened.
When I set header_len + 5 (9), we can avoid the problem.
As far as I looked at the test bz2 file using binary editor, the first bz2
marker starts from position 4 (right after bz2 header).
Still trying to understand why we need to set header_len + 5.
> BZip2 drops and duplicates records when input split size is small
> -----------------------------------------------------------------
>
> Key: HADOOP-15206
> URL: https://issues.apache.org/jira/browse/HADOOP-15206
> Project: Hadoop Common
> Issue Type: Bug
> Affects Versions: 2.8.3, 3.0.0
> Reporter: Aki Tanaka
> Priority: Major
> Attachments: HADOOP-15206-test.patch, HADOOP-15206.001.patch,
> HADOOP-15206.002.patch
>
>
> BZip2 can drop and duplicate record when input split file is small. I
> confirmed that this issue happens when the input split size is between 1byte
> and 4bytes.
> I am seeing the following 2 problem behaviors.
>
> 1. Drop record:
> BZip2 skips the first record in the input file when the input split size is
> small
>
> Set the split size to 3 and tested to load 100 records (0, 1, 2..99)
> {code:java}
> 2018-02-01 10:52:33,502 INFO [Thread-17] mapred.TestTextInputFormat
> (TestTextInputFormat.java:verifyPartitions(317)) -
> splits[1]=file:/work/count-mismatch2/hadoop/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/target/test-dir/TestTextInputFormat/test.bz2:3+3
> count=99{code}
> > The input format read only 99 records but not 100 records
>
> 2. Duplicate Record:
> 2 input splits has same BZip2 records when the input split size is small
>
> Set the split size to 1 and tested to load 100 records (0, 1, 2..99)
>
> {code:java}
> 2018-02-01 11:18:49,309 INFO [Thread-17] mapred.TestTextInputFormat
> (TestTextInputFormat.java:verifyPartitions(318)) - splits[3]=file
> /work/count-mismatch2/hadoop/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/target/test-dir/TestTextInputFormat/test.bz2:3+1
> count=99
> 2018-02-01 11:18:49,310 WARN [Thread-17] mapred.TestTextInputFormat
> (TestTextInputFormat.java:verifyPartitions(308)) - conflict with 1 in split 4
> at position 8
> {code}
>
> I experienced this error when I execute Spark (SparkSQL) job under the
> following conditions:
> * The file size of the input files are small (around 1KB)
> * Hadoop cluster has many slave nodes (able to launch many executor tasks)
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org
