CODEWITHSUNDEEP
amazon-web-servicesamazon-redshiftaws-glue
vtbiggi
vtbiggi

Reputation: 183

Timestamp not Loading into Redshift Table from Glue

I have a timestamp in YYYY-MM-DD XX:XX:XX format in a csv file that is stored in S3 but when I use the timestamp data type to load into a Redshift database using Glue, the timestamp column is null. It appears that format is valid but I've also tried YYYYMMDD XXXXXX and YYMMDD XX:XX:XX formats as well just incase.

My mapping in Glue goes from timestamp to timestamp and the column datatype of the table is also timestamp. Ex of data in csv format:

1,2016 Summer,2016-06-22 00:00:00

Actual Output:

Line |    Term     |    Date
-----+-------------+------------
   1 | 2016 Summer |

Expected Output:

Line |    Term     |        Date
-----+-------------+---------------------
   1 | 2016 Summer | 2016-06-22 00:00:00

It seems that this should be a straightforward task but I can't get it right so if anyone else can find my mistake(s), that would be greatly appreciated.

Code:

val datasource37 = glueContext.getCatalogSource(database = "data", tableName = "term", redshiftTmpDir = "", transformationContext = "datasource37").getDynamicFrame()
val applymapping37 = datasource37.applyMapping(mappings = Seq(("id", "bigint", "id", "bigint"), ("name", "string", "name", "varchar(256)"), ("date", "timestamp", "date_start", "timestamp")), caseSensitive = false, transformationContext = "applymapping37")
val resolvechoice37 = applymapping37.resolveChoice(choiceOption = Some(ChoiceOption("make_cols")), transformationContext = "resolvechoice37")
val dropnullfields37 = resolvechoice37.dropNulls(transformationContext = "dropnullfields37")
val datasink37 = glueContext.getJDBCSink(catalogConnection = "dataConnection", options = JsonOptions("""{"dbtable": "term", "database": "data"}"""), redshiftTmpDir = args("TempDir"), transformationContext = "datasink37").writeDynamicFrame(dropnullfields37)

Upvotes: 0

Views: 1347

Answers (1)

vtbiggi
vtbiggi

Reputation: 183

I ended up mapping from string -> timestamp and it worked. Glue had it automatically mapping from timestamp -> timestamp so I assumed it was right.

Ex:

val applymapping37 = datasource37.applyMapping
                     (mappings = Seq(("id", "bigint", "id", "bigint"),
                     ("name", "string", "name", "varchar(256)"), 
                     ("date", "string", "date_start", "timestamp")),
                     caseSensitive = false, transformationContext = "applymapping37")

Upvotes: 5

Related Questions