Make integration tests runnable in distributed mode
- 主要语言
- Scala
- 星标
- 33
- 派生
- 16
- PR 合并指标
- 30 天内没有已合并 PR
描述
## Background
Most of the unit tests can be ran in distributed mode on a Hadoop cluster already.
But some integration tests rely on input files located in a local file system to be available to a Spark application. Since cluster configuration uses HDFS instead of local filesystem for storing files, Standardization and Conformance jobs cannot find expected files and fail.
## Feature
Make Standardization and Conformance integration tests runnable in distributed mode on a cluster.
## Proposed Solution
Here is an idea for a possible solution.
In order for Conformance integration tests to work on a cluster:
* A path to a temporary HDFS folder should be provided in `application.conf`. The folder should have read/write access.
* Input tests data and mapping tables should be generated before any integration test is ran. The files should be put to the specified temporary folder.
* Integration tests should use the configured temporary path location in all dataset and mapping table definitions.
贡献指南
评估
这个 Issue 还没有评估数据。