You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@beam.apache.org by "yangping wu (JIRA)" <ji...@apache.org> on 2017/04/01 06:11:41 UTC
[jira] [Updated] (BEAM-1491) HDFSFileSource should be able to read
the HADOOP_CONF_DIR(YARN_CONF_DIR) environmen variable
[ https://issues.apache.org/jira/browse/BEAM-1491?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
yangping wu updated BEAM-1491:
------------------------------
Summary: HDFSFileSource should be able to read the HADOOP_CONF_DIR(YARN_CONF_DIR) environmen variable (was: HDFSFileSource should be able to read the HADOOP_CONF(YARN_CONF) environmen variable)
> HDFSFileSource should be able to read the HADOOP_CONF_DIR(YARN_CONF_DIR) environmen variable
> --------------------------------------------------------------------------------------------
>
> Key: BEAM-1491
> URL: https://issues.apache.org/jira/browse/BEAM-1491
> Project: Beam
> Issue Type: Improvement
> Components: sdk-java-core
> Affects Versions: 0.5.0
> Reporter: yangping wu
> Assignee: Jean-Baptiste Onofré
>
> Currently, if we want to read file store on HDFS, we will do it as follow:
> {code} PCollection<KV<LongWritable, Text>> resultCollection = p.apply(HDFSFileSource.readFrom(
> "hdfs://hadoopserver:8020/tmp/data.txt",
> TextInputFormat.class, LongWritable.class, Text.class));
> {code}
> As we have seen above, we must be set {{hdfs://hadoopserver:8020}} in the file path, and we cann't set any variables when read file, because in [HDFSFileSource.java|https://github.com/apache/beam/blob/master/sdks/java/io/hdfs/src/main/java/org/apache/beam/sdk/io/hdfs/HDFSFileSource.java#L310] we initialize {{job}} instance as follow:
> {code}
> this.job = Job.getInstance();
> {code}
> we should initialize {{job}} instance by configure:
> {code}
> this.job = Job.getInstance(conf);
> {code}
> where {{conf}} is instance of {{Configuration}}, and we initialize {{conf}} by reading {{HADOOP_CONF}}({{YARN_CONF}}) environmen variable,then we can read HDFS file like this:
> {code} PCollection<KV<LongWritable, Text>> resultCollection = p.apply(HDFSFileSource.readFrom(
> "/tmp/data.txt",
> TextInputFormat.class, LongWritable.class, Text.class));
> {code}
> note we don't specify {{hdfs://hadoopserver:8020}} prefix, because the program read it from {{HADOOP_CONF}}({{YARN_CONF}}) environmen.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)