Lambda Function for Signalk Batch Format to Timestream

This is the lambda function I use for processing the batch jsons uploaded to S3 by signalk-to-batch-format and then writing to Amazon Timestream.

See signalk-to-batch-format for more detail on the setup and motivation.

Setup

First, do an npm install in this directory, to get all the dependencies. This might be cleanest in a docker instance:

docker run -it -v $(pwd):/code node:12 bash
cd /code
npm install

Then, from outside docker, and with proper AWS lambda credentials configured, just run make:

make

Note, that if you want to name your lambda something other than signalk-to-timestream, edit the Makefile.

The lambda you create needs a role with the following permissions:

• S3:GetObject
• Timestream:DescribeEndpoints
• Timestream:WriteRecords
• Logs:CreateLogGroup
• Logs:CreateLogStream
• Logs:PutLogEvents

The logs permissions are optional.

In the lambda, set the following environment variables:

• DATABASE_NAME Your Timestream Database Name
• TABLE_NAME Your Timestream Table Name
• SELF_ID Your self id string from signalk (ie: urn:mrn:imo:mmsi:1234)
• DELETE_FROM_S3_ON_SUCCESS (Optional) the string "yes" or "no" to indicate whether this script should do the delete from S3

Depending on how many metrics you're publishing to timestream, you may need to slightly increase the RAM and/or timeout of your lambda. I'm running with 192MB of RAM and a 10s timeout.