|
| 1 | +package dynamo |
| 2 | + |
| 3 | +import ( |
| 4 | + "github.com/aws/aws-sdk-go/aws" |
| 5 | + "github.com/aws/aws-sdk-go/service/dynamodb" |
| 6 | + "github.com/halprin/delete-dynamodb-items/parallel" |
| 7 | + "log" |
| 8 | + "time" |
| 9 | +) |
| 10 | + |
| 11 | +var maxItemsPerBatchRequest = 25 |
| 12 | +var tableKeys []*dynamodb.KeySchemaElement |
| 13 | + |
| 14 | +func deleteItems(dynamoItems []map[string]*dynamodb.AttributeValue, tableName string) error { |
| 15 | + |
| 16 | + var err error |
| 17 | + tableKeys, err = getTableKeys(tableName) |
| 18 | + if err != nil { |
| 19 | + log.Println("Unable to determine the keys of the table") |
| 20 | + return err |
| 21 | + } |
| 22 | + |
| 23 | + dynamoItemsChunks := chunkItems(dynamoItems) |
| 24 | + |
| 25 | + var errorChannels []chan error |
| 26 | + |
| 27 | + for _, currentItemsChunk := range dynamoItemsChunks { |
| 28 | + errorChannel := make(chan error) |
| 29 | + errorChannels = append(errorChannels, errorChannel) |
| 30 | + go deleteChunkGoroutine(currentItemsChunk, tableName, errorChannel) |
| 31 | + } |
| 32 | + |
| 33 | + for errorFromGoroutine := range parallel.MergeErrorChannels(errorChannels) { |
| 34 | + if errorFromGoroutine != nil { |
| 35 | + log.Println("One of the delete goroutines failed") |
| 36 | + return errorFromGoroutine |
| 37 | + } |
| 38 | + } |
| 39 | + |
| 40 | + return nil |
| 41 | +} |
| 42 | + |
| 43 | +func deleteChunkGoroutine(currentItemsChunk []map[string]*dynamodb.AttributeValue, tableName string, errorChannel chan error) { |
| 44 | + errorChannel <- deleteChunk(currentItemsChunk, tableName) |
| 45 | + close(errorChannel) |
| 46 | +} |
| 47 | + |
| 48 | +func deleteChunk(currentItemsChunk []map[string]*dynamodb.AttributeValue, tableName string) error { |
| 49 | + writeRequests := marshalItemsIntoBatchWrites(currentItemsChunk) |
| 50 | + |
| 51 | + requestItems := map[string][]*dynamodb.WriteRequest{ |
| 52 | + tableName: writeRequests, |
| 53 | + } |
| 54 | + |
| 55 | + err := incrementallyBatchDelete(requestItems) |
| 56 | + if err != nil { |
| 57 | + log.Println("Failed to batch delete items") |
| 58 | + return err |
| 59 | + } |
| 60 | + |
| 61 | + return nil |
| 62 | +} |
| 63 | + |
| 64 | +func getTableKeys(tableName string) ([]*dynamodb.KeySchemaElement, error) { |
| 65 | + describeTableInput := &dynamodb.DescribeTableInput{ |
| 66 | + TableName: aws.String(tableName), |
| 67 | + } |
| 68 | + |
| 69 | + tableInfo, err := dynamoService.DescribeTable(describeTableInput) |
| 70 | + if err != nil { |
| 71 | + log.Println("Unable to describe the the table") |
| 72 | + return nil, err |
| 73 | + } |
| 74 | + |
| 75 | + return tableInfo.Table.KeySchema, nil |
| 76 | +} |
| 77 | + |
| 78 | +func chunkItems(dynamoItems []map[string]*dynamodb.AttributeValue) [][]map[string]*dynamodb.AttributeValue { |
| 79 | + var itemChunks [][]map[string]*dynamodb.AttributeValue |
| 80 | + numberOfItems := len(dynamoItems) |
| 81 | + |
| 82 | + for itemIndex := 0; itemIndex < numberOfItems; itemIndex += maxItemsPerBatchRequest { |
| 83 | + end := itemIndex + maxItemsPerBatchRequest |
| 84 | + |
| 85 | + if end > numberOfItems { |
| 86 | + end = numberOfItems |
| 87 | + } |
| 88 | + |
| 89 | + itemChunks = append(itemChunks, dynamoItems[itemIndex:end]) |
| 90 | + } |
| 91 | + |
| 92 | + return itemChunks |
| 93 | +} |
| 94 | + |
| 95 | +func marshalItemsIntoBatchWrites(dynamoItems []map[string]*dynamodb.AttributeValue) []*dynamodb.WriteRequest { |
| 96 | + var writeRequests []*dynamodb.WriteRequest |
| 97 | + var writeRequest *dynamodb.WriteRequest |
| 98 | + |
| 99 | + for _, currentDynamoItem := range dynamoItems { |
| 100 | + key := convertItemToKey(currentDynamoItem) |
| 101 | + |
| 102 | + deleteRequest := &dynamodb.DeleteRequest{ |
| 103 | + Key: key, |
| 104 | + } |
| 105 | + |
| 106 | + writeRequest = &dynamodb.WriteRequest{ |
| 107 | + DeleteRequest: deleteRequest, |
| 108 | + } |
| 109 | + |
| 110 | + writeRequests = append(writeRequests, writeRequest) |
| 111 | + } |
| 112 | + |
| 113 | + return writeRequests |
| 114 | +} |
| 115 | + |
| 116 | +func convertItemToKey(item map[string]*dynamodb.AttributeValue) map[string]*dynamodb.AttributeValue { |
| 117 | + key := make(map[string]*dynamodb.AttributeValue) |
| 118 | + for _, currentTableKey := range tableKeys { |
| 119 | + currentTableKeyName := *currentTableKey.AttributeName |
| 120 | + key[currentTableKeyName] = item[currentTableKeyName] |
| 121 | + } |
| 122 | + |
| 123 | + return key |
| 124 | +} |
| 125 | + |
| 126 | +func incrementallyBatchDelete(requestItems map[string][]*dynamodb.WriteRequest) error { |
| 127 | + millisecondsToWait := 20 |
| 128 | + |
| 129 | + for { |
| 130 | + batchWriteItemInput := &dynamodb.BatchWriteItemInput{ |
| 131 | + RequestItems: requestItems, |
| 132 | + } |
| 133 | + |
| 134 | + log.Println("Deleting some items") |
| 135 | + |
| 136 | + batchWriteItemOutput, err := dynamoService.BatchWriteItem(batchWriteItemInput) |
| 137 | + if err != nil { |
| 138 | + //there was an error writing to DynamoDB |
| 139 | + log.Println("Failed to put/delete items in DynamoDB") |
| 140 | + return err |
| 141 | + } |
| 142 | + |
| 143 | + if len(batchWriteItemOutput.UnprocessedItems) > 0 { |
| 144 | + //there are still items to write, reset requestItems for the next pass |
| 145 | + log.Println("Unprocessed items remain, trying again with remaining items") |
| 146 | + requestItems = batchWriteItemOutput.UnprocessedItems |
| 147 | + } else { |
| 148 | + //no more items to write, break out |
| 149 | + break |
| 150 | + } |
| 151 | + |
| 152 | + //do an exponential back-off |
| 153 | + time.Sleep(time.Duration(millisecondsToWait) * time.Millisecond) |
| 154 | + millisecondsToWait *= 2 |
| 155 | + } |
| 156 | + |
| 157 | + return nil |
| 158 | +} |
0 commit comments