使用带有分区键和排序键的bash删除DynamoDB表中的所有项

Question

我正在尝试使用bash中的AWS CLI删除同时具有分区和排序键的DynamoDB表中的所有项目。到目前为止，我发现的最好的事情是：

aws dynamodb scan --table-name $TABLE_NAME --attributes-to-get "$KEY" \
--query "Items[].$KEY.S" --output text | \
tr "\t" "\n" | \
xargs -t -I keyvalue aws dynamodb delete-item --table-name $TABLE_NAME \
--key "{\"$KEY\": {\"S\": \"keyvalue\"}}"

但这对于同时具有分区键和排序键的表不起作用，而且我还不能使它对这样的表起作用。知道如何修改脚本以使其适用于具有复合键的表吗？

Answer 1

如果您对使用Node.js感兴趣，请查看此示例（我在这里使用TypeScript）。更多相关信息可以在AWS文档中找到。

import AWS from 'aws-sdk';
const DynamoDb = new AWS.DynamoDB.DocumentClient({
region: 'eu-west-1'

});
export const getAllItemsFromTable = async TableName => {
   const Res = await DynamoDb.scan({ TableName }).promise();
   return Res.Items;
};

export const deleteAllItemsFromTable = async (TableName = '', items:{ id: string }, hashKey) => {
  var counter = 0;
  //split items into patches of 25
  // 25 items is max for batchWrite
  asyncForEach(split(items, 25), async (patch, i) => {
    const RequestItems = {
      TableName: patch.map(item => {
        return {
          DeleteRequest: {
            Key: {
              id: item.id
            }
          }
        };
      })
    };
    await DynamoDb.batchWrite({ RequestItems }).promise();
    counter += patch.length;
    console.log('counter : ', counter);
  });
};

function split(arr, n) {
  var res = [];
  while (arr.length) {
    res.push(arr.splice(0, n));
  }
  return res;
}

async function asyncForEach(array, callback) {
  for (let index = 0; index < array.length; index++) {
    await callback(array[index], index, array);
  }
}

const tableName = "table"
// assuming table hashKey is named "id"
deleteAllItemsFromTable(tableName,getAllItemsFromTable(tableName))

Answer 2

更正@Cheruvian发布的内容。以下命令有效，创建架构时需要排除的字段很少。json。

aws dynamodb describe-table --table-name $table_name | jq '.Table | del(.TableId, .TableArn, .ItemCount, .TableSizeBytes, .CreationDateTime, .TableStatus, .LatestStreamArn, .LatestStreamLabel, .ProvisionedThroughput.NumberOfDecreasesToday, .ProvisionedThroughput.LastIncreaseDateTime)' > schema.json

aws dynamodb delete-table --table-name $table_name

aws dynamodb create-table --cli-input-json file://schema.json

Answer 3

根据表的大小，这可能会太贵，导致停机。请记住，删除的成本与写入的成本相同，因此您将受到配置的WCU的限制。只需删除并重新创建表就可以更简单、更快。

# this uses jq but basically we're just removing 
# some of the json fields that describe an existing 
# ddb table and are not actually part of the table schema/defintion
aws dynamodb describe-table --table-name $table_name | jq '.Table | del(.TableId, .TableArn, .ItemCount, .TableSizeBytes, .CreationDateTime, .TableStatus, .ProvisionedThroughput.NumberOfDecreasesToday)' > schema.json
# delete the table
aws dynamodb delete-table --table-name $table_name
# create table with same schema (including name and provisioned capacity)
aws dynamodb create-table --cli-input-json file://schema.json

如果您真的愿意，您可以单独删除每个项目，并且您走在正确的轨道上，您只需要在扫描投影和删除命令中指定散列和范围键。

aws dynamodb scan \
  --attributes-to-get $HASH_KEY $RANGE_KEY \
  --table-name $TABLE_NAME --query "Items[*]" \
  # use jq to get each item on its own line
  | jq --compact-output '.[]' \
  # replace newlines with null terminated so 
  # we can tell xargs to ignore special characters 
  | tr '\n' '\0' \
  | xargs -0 -t -I keyItem \
    # use the whole item as the key to delete (dynamo keys *are* dynamo items)
    aws dynamodb delete-item --table-name $TABLE_NAME --key=keyItem

如果你想得到超级花哨的，你可以使用state-table调用来获取散列和范围键来填充$HASH_KEY和$RANGE_KEY，但我会把它作为一个练习留给你。