Got it, let's walk through how to delete those items efficiently. The core challenge here is taking the results from your scan command and feeding them into delete operations—since AWS CLI doesn't have a built-in "delete from scan" command, we'll use a combination of jq (for parsing JSON) and loop/batch operations to get this done.

Step 1: Validate Your Scan Command First #

First, make sure your scan is correctly returning the items you want to delete. Note that --scan-filter is a deprecated parameter—AWS recommends using --filter-expression instead for better compatibility. Here's the updated command (replace id with your actual primary key name if it's different):

# Using filter-expression (recommended)
aws dynamodb scan --table-name test --filter-expression "attribute_exists(demo)" --output json

If you still want to use your filter.json file, the command would be:

aws dynamodb scan --table-name test --scan-filter file://D:\filter.json --output json

Run this command and verify that the Items array contains exactly the entries you want to delete. Important: Ensure the output includes your table's primary key(s)—you'll need these to delete items.

Step 2: Delete Items One by One (Simple Approach) #

If you don't have thousands of items, a simple loop works. We'll use jq to extract the primary key from each scan result, then pass it to delete-item.

For Bash/Linux/Mac: #

# Replace "id" with your actual primary key name
aws dynamodb scan --table-name test --filter-expression "attribute_exists(demo)" --output json | \
jq -c '.Items[] | {Key: {id: .id}}' | \
while read -r item_key; do
  aws dynamodb delete-item --table-name test --key "$item_key"
done

For Windows PowerShell: #

# Replace "id" with your actual primary key name
$scanResults = aws dynamodb scan --table-name test --filter-expression "attribute_exists(demo)" --output json | ConvertFrom-Json
foreach ($item in $scanResults.Items) {
  $keyJson = @{ id = $item.id } | ConvertTo-Json -Compress
  aws dynamodb delete-item --table-name test --key $keyJson
}

Step 3: Batch Delete (For Large Datasets) #

If you have hundreds or thousands of items, single delete-item calls will be slow. Use batch-write-item instead—it lets you delete up to 25 items per request, which is much more efficient.

Bash/Linux/Mac Example: #

# 1. Extract all delete requests (replace "id" with your primary key)
aws dynamodb scan --table-name test --filter-expression "attribute_exists(demo)" --output json | \
jq -c '.Items[] | {DeleteRequest: {Key: {id: .id}}}' > delete_requests.json

# 2. Split into batches of 25
split -l 25 delete_requests.json batch_

# 3. Process each batch
for batch_file in batch_*; do
  # Convert the batch into the required format for batch-write-item
  batch_content=$(jq -s '.' "$batch_file")
  aws dynamodb batch-write-item --request-items "{\"test\": $batch_content}"
  rm "$batch_file"
done

rm delete_requests.json

Critical Notes to Avoid Mistakes #

• Handle Pagination: If your scan returns more than 1MB of data (DynamoDB's default scan limit), you'll need to use the --starting-token parameter or AWS CLI's paginator to fetch all items. Add --paginate to your scan command to auto-handle pagination:

  aws dynamodb paginate scan --table-name test --filter-expression "attribute_exists(demo)" --output json
  

• Backup First: Always export the scan results to a backup file before deleting, just in case:

  aws dynamodb scan --table-name test --filter-expression "attribute_exists(demo)" --output json > backup_demo_items.json
  

• Match Primary Key Types: If your primary key is a number (not a string), adjust the jq command to use .id.N instead of .id.S (or just .id if you're letting AWS CLI infer the type).
• Rate Limits: Be aware of DynamoDB's write capacity limits—if you're deleting a lot of items, you might need to throttle your requests or adjust your table's provisioned capacity temporarily.

内容的提问来源于stack exchange，提问作者Jayendran