[
https://issues.apache.org/jira/browse/HIVE-14792?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16393380#comment-16393380
]
Mithun Radhakrishnan commented on HIVE-14792:
---------------------------------------------
Ok, this patch does not look ready to go in. :/ I was hoping these were
transient failures, but it looks like [^HIVE-14792.3.patch] is still failing
hard, based on the test-results above.
[^HIVE-14792.4.patch] happens to run, because the optimization is not enabled.
This is back to where it was when I ran out of time. :[ I'll update if I can
take a crack at this.
> AvroSerde reads the remote schema-file at least once per mapper, per table
> reference.
> -------------------------------------------------------------------------------------
>
> Key: HIVE-14792
> URL: https://issues.apache.org/jira/browse/HIVE-14792
> Project: Hive
> Issue Type: Bug
> Affects Versions: 1.2.1, 2.1.0
> Reporter: Mithun Radhakrishnan
> Assignee: Aihua Xu
> Priority: Major
> Labels: TODOC2.2, TODOC2.4
> Fix For: 3.0.0, 2.4.0, 2.2.1
>
> Attachments: HIVE-14792.1.patch, HIVE-14792.3.patch,
> HIVE-14792.4.patch
>
>
> Avro tables that use "external" schema files stored on HDFS can cause
> excessive calls to {{FileSystem::open()}}, especially for queries that spawn
> large numbers of mappers.
> This is because of the following code in {{AvroSerDe::initialize()}}:
> {code:title=AvroSerDe.java|borderStyle=solid}
> public void initialize(Configuration configuration, Properties properties)
> throws SerDeException {
> // ...
> if (hasExternalSchema(properties)
> || columnNameProperty == null || columnNameProperty.isEmpty()
> || columnTypeProperty == null || columnTypeProperty.isEmpty()) {
> schema = determineSchemaOrReturnErrorSchema(configuration, properties);
> } else {
> // Get column names and sort order
> columnNames = Arrays.asList(columnNameProperty.split(","));
> columnTypes =
> TypeInfoUtils.getTypeInfosFromTypeString(columnTypeProperty);
> schema = getSchemaFromCols(properties, columnNames, columnTypes,
> columnCommentProperty);
>
> properties.setProperty(AvroSerdeUtils.AvroTableProperties.SCHEMA_LITERAL.getPropName(),
> schema.toString());
> }
> // ...
> }
> {code}
> For tables using {{avro.schema.url}}, every time the SerDe is initialized
> (i.e. at least once per mapper), the schema file is read remotely. For
> queries with thousands of mappers, this leads to a stampede to the handful
> (3?) datanodes that host the schema-file. In the best case, this causes
> slowdowns.
> It would be preferable to distribute the Avro-schema to all mappers as part
> of the job-conf. The alternatives aren't exactly appealing:
> # One can't rely solely on the {{column.list.types}} stored in the Hive
> metastore. (HIVE-14789).
> # {{avro.schema.literal}} might not always be usable, because of the
> size-limit on table-parameters. The typical size of the Avro-schema file is
> between 0.5-3MB, in my limited experience. Bumping the max table-parameter
> size isn't a great solution.
> If the {{avro.schema.file}} were read during query-planning, and made
> available as part of table-properties (but not serialized into the
> metastore), the downstream logic will remain largely intact. I have a patch
> that does this.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
