问题描述
场景：
我使用BulkExecutor库进行批量插入。
我使用了Github文档中提到的以下设置，将所有控件传递给BulkExecutor库。
//将重试设置为0以通过完全控制批量执行程序。
client.ConnectionPolicy.RetryOptions。 MaxRetryWaitTimeInSeconds = 0;
client.ConnectionPolicy.RetryOptions。 MaxRetryAttemptsOnThrottledRequests = 0;
$
我正在插入 如下所示，请注意 enableUpsert 和 disableAutomaticIdGeneration 设置为
false 。
try
{
var bulkImportResponse = await bulkExecutor.BulkImportAsync（documents：documentsToImportInBatch，enableUpsert：false，disableAutomaticIdGeneration：false）;
}
catch（例外e）
{
//错误处理逻辑
}
案例：
Scenario:
I am using BulkExecutor library for Bulk insertion.
I have used following settings as per mentioned in Github docs, to pass all control to BulkExecutor library.
// Set retries to 0 to pass complete control to bulk executor.
client.ConnectionPolicy.RetryOptions.MaxRetryWaitTimeInSeconds = 0;
client.ConnectionPolicy.RetryOptions.MaxRetryAttemptsOnThrottledRequests = 0;
I am inserting as given below, note that enableUpsert and disableAutomaticIdGeneration are set tofalse.
try
{
var bulkImportResponse = await bulkExecutor.BulkImportAsync(documents: documentsToImportInBatch, enableUpsert: false, disableAutomaticIdGeneration: false);
}
catch (Exception e)
{
// error handling logic
}
Case:
假设我已经传递了100个文件用于批量插入，并且在插入60个文档后发生了异常。
Suppose I have passed 100 documents for bulk insertion and exception occurs after insertion of 60 documents.
问题：
1。 bulkExecutor是否回滚插入操作。
2.如果否，那么仅重新插入40个文档的机制是什么。
3.我怎么知道哪40个文件是未被BulkExecutor插入。
如果您遇到同样的问题，我想知道任何其他解决方案。
Questions:
1. Does bulkExecutor rollback insertion operation.
2. If No, then What is the mechanism to retry insertion for 40 documents only.
3. How do I know which 40 documents are not inserted by BulkExecutor.
I would like to know any other solutions if you have related to same issue.
推荐答案
批量执行程序不回滚，但Cosmosdb通常会返回服务器异常，例如服务器不接受请求。通常我们创建存储过程以进行回滚操作。
Bulk executor does not rollback but Cosmosdb typically returns server exception somthing like server not accepting the request. Typically we create Stored procedure to do have a roll back action.
如果您有大量负载插入，您可以检查这个
You can go over this if you have heavy load insertions
https：// stackoverflow.com/questions/41744582/fastest-way-to-insert-100-000-records-into-documentdb
https://stackoverflow.com/questions/41744582/fastest-way-to-insert-100-000-records-into-documentdb
另外，如果你想跳过存储过程，你可以实现客户端模式将发送取消令牌以检查插入的记录。
Also, incase you want to skip the stored procedure, you can implement a client side pattern that will send cancellation tokens accross to check the inserted records.
有几个博客列出了这些，我觉得这很有用： Cosmos DB存储过程 - 处理延续
There are couple of blogs which list those, I find this useful: Cosmos DB Stored Procedures – handling continuation
干杯
K
这篇关于如何在使用BulkExecutor库时重试批量插入失败。的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！