如何为 Spark Java 提供动态数据库配置凭据？

Question

我正在使用 spark java 构建一个简单的 REST API。该项目使用 Postgres 数据库进行存储，gradle 作为构建系统。 fat Jar 部署在 AWS.

我有三个环境 local、staging 和 production。对于每个 fat jar 创建，我想提供如下用于本地构建的特定配置：

    dbHost = "localhost"
    dbUsername = "myusername"
    dbPassword = "mypassword"
    database = "mydb"
    dbPort = "5432"

如何根据 fatJar 构建变体动态提供这些值？

Answer 1

我按照这个示例 github 项目 Kotlin-dev-proxy.

使用了 .properties 文件

服务器读取名为 app_environment 的环境变量并使用适当的资源文件。

示例代码：

DbConfig.kt

data class DbConfig(
        val dbHost: String,
        val dbUsername: String,
        val dbPassword: String,
        val dbName: String,
        val dbPort: Int)

ServerSettings.kt

class ServerSettings(settings: String) {
    val resources = Properties()

    init {
        val fileUrl: URL = resources.javaClass.getResource("/$settings.properties")
                ?: throw FileNotFoundException("$settings.properties file not found")
        fileUrl.openStream().use { resources.load(it) }
    }

    fun printSettings() = resources.stringPropertyNames().forEach {
        println("Property: $it has value: '${resources[it]}'")
    }

    fun getString(key: String): String = resources[key]!! as String

    fun getInt(key: String): Int = (resources[key]!! as String).toInt()
}

DbConfig.kt

fun provideDbConfig(): DbConfig {
        val settings = ServerSettings(System.getenv("app_environment"))
        settings.printSettings()

        return DbConfig(
                dbHost = settings.getString("dbHost"),
                dbUsername = settings.getString("dbUsername"),
                dbPassword = settings.getString("dbPassword"),
                dbName = settings.getString("dbName"),
                dbPort = settings.getInt("dbPort")
        )
    }

resources/develop.properties

# Local Db values
dbHost=localhost
dbUsername=postgresusername
dbPassword=postgrespassword
dbName=mydbname
dbPort=5432

在 Amazon EBS 中部署时，转到软件配置并为环境添加适当的环境变量(staging/production)。

如何为 Spark Java 提供动态数据库配置凭据？

How to provide dynamic database configuration credentials for Spark Java?

java

gradle

kotlin

spark-java