How it works

ddb-data-check performs table schema checks and item-level comparisons. It does not call a write API.

Schema check

The command calls DescribeTable on source and target and checks the partition key name, optional sort key name, and each local secondary index name and sort key name. A mismatch is written as table_schema_diff, and item checking stops for that table.

Data comparison pipeline

check_mode selects the baseline. Source mode scans source; target mode scans target.

Scan the baseline endpoint
    ↓
Extract the primary key and full item
    ↓
GetItem from the peer endpoint
    ↓
Deserialize and compare fields
    ↓
Write the JSON difference report

The original primary-key AttributeValue from the Scan response is used directly for peer GetItem, preserving numeric key representations.

Item comparison

Both items are deserialized to map[string]interface{}. The command first normalizes numeric types and compares the maps. If they differ, it recursively sorts slices and compares again to handle unordered DynamoDB Sets.

Results include:

• item_missing_in_peer: the peer endpoint has no item for the key.
• item_diff: both endpoints have the item, but its fields differ.
• item_error: GetItem, deserialization, or another read operation failed.
• table_summary: compared, diff, missing, and error totals.

Direction and consistency boundary

check_mode = "source" does not detect target-only items. Run again with check_mode = "target" when they must also be checked.

Scan and GetItem do not share a snapshot timestamp. Application writes during the check can appear as temporary differences.