Apache
/
druid
mirror of https://github.com/apache/druid.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Perform lazy initialization of parquet extensions module (#12827) 

Historicals and middle managers crash with an `UnknownHostException` on trying
to load `druid-parquet-extensions` with an ephemeral Hadoop cluster. This happens
because the `fs.defaultFS` URI value cannot be resolved at start up time as the
hadoop cluster may not exist at startup time.

This commit fixes the error by performing initialization of the filesystem in
`ParquetInputFormat.createReader()` whenever a new reader is requested.
This commit is contained in:
Tejaswini Bandlamudi 2022-08-02 13:41:12 +05:30 committed by GitHub
parent 6046a392b6
commit cceb2e849e
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 23 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
extensions-core/parquet-extensions/src/main/java/org/apache/druid/data/input/parquet/ParquetExtensionsModule.java
View File

@ -30,9 +30,7 @@ import org.apache.druid.data.input.parquet.simple.ParquetHadoopInputRowParser;
import org.apache.druid.data.input.parquet.simple.ParquetParseSpec;
import org.apache.druid.initialization.DruidModule;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import java.io.IOException;
import java.util.Collections;
import java.util.List;
import java.util.Properties;
@ -77,20 +75,6 @@ public class ParquetExtensionsModule implements DruidModule
// Set explicit CL. Otherwise it'll try to use thread context CL, which may not have all of our dependencies.
conf.setClassLoader(getClass().getClassLoader());
// Ensure that FileSystem class level initialization happens with correct CL
// See https://github.com/apache/druid/issues/1714
ClassLoader currCtxCl = Thread.currentThread().getContextClassLoader();
try {
Thread.currentThread().setContextClassLoader(getClass().getClassLoader());
FileSystem.get(conf);
}
catch (IOException ex) {
throw new RuntimeException(ex);
}
finally {
Thread.currentThread().setContextClassLoader(currCtxCl);
}
if (props != null) {
for (String propName : props.stringPropertyNames()) {
if (propName.startsWith("hadoop.")) {

23
extensions-core/parquet-extensions/src/main/java/org/apache/druid/data/input/parquet/ParquetInputFormat.java
View File

@ -29,9 +29,11 @@ import org.apache.druid.data.input.impl.NestedInputFormat;
import org.apache.druid.data.input.parquet.guice.Parquet;
import org.apache.druid.java.util.common.parsers.JSONPathSpec;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import javax.annotation.Nullable;
import java.io.File;
import java.io.IOException;
import java.util.Objects;
public class ParquetInputFormat extends NestedInputFormat
@ -51,6 +53,26 @@ public class ParquetInputFormat extends NestedInputFormat
this.conf = conf;
}
private void initialize(Configuration conf)
{
// Initializing seperately since during eager initialization, resolving
// namenode hostname throws an error if nodes are ephemeral
// Ensure that FileSystem class level initialization happens with correct CL
// See https://github.com/apache/druid/issues/1714
ClassLoader currCtxCl = Thread.currentThread().getContextClassLoader();
try {
Thread.currentThread().setContextClassLoader(getClass().getClassLoader());
FileSystem.get(conf);
}
catch (IOException ex) {
throw new RuntimeException(ex);
}
finally {
Thread.currentThread().setContextClassLoader(currCtxCl);
}
}
@JsonProperty
public boolean getBinaryAsString()
{
@ -70,6 +92,7 @@ public class ParquetInputFormat extends NestedInputFormat
File temporaryDirectory
)
{
initialize(conf);
return new ParquetReader(conf, inputRowSchema, source, temporaryDirectory, getFlattenSpec(), binaryAsString);
}