How do I specify datatypes in Timestream multi-measure record writes?

[nabelekt]

Hello. I am using awswrangler.timestream.write (https://aws-sdk-pandas.readthedocs.io/en/stable/stubs/awswrangler.timestream.write.html) to write multi-measuare records to Timestream, similar to how it's being done here: https://github.com/aws/aws-sdk-pandas/blob/main/tests/unit/test_timestream.py#L378. But, all of the values other than time are stored with the type VARCHAR. I need some to be BIGINT. How can I do this? I know that I can do it if I write the records using boto3 directly, but now I am trying to do it from a Pandas dataframe. Thanks.

[nabelekt]

@jaidisido, @kukushking, or @LeonLuttenberger, can you help here? Thanks.

[LeonLuttenberger]

Hey,

Can you provide an example of a dataframe where the columns are returned as strings? I ran the example from our unit tests, and measure1 and measure2 are returned as float values, since they're specified as measure columns.

The values that are getting converted into strings are the dimensions (dim0 and dim1). Based on the documentation, it looks like they have to be strings unfortunately.

[nabelekt]

Indeed, I think all of the non-string values I was testing were being used as dimensions. Doing it differently, I have it working as expected. Thanks.