This is a fork of Allan Helton's project here which was developed for use with AWS SAM. SST is my personal favourite means of deploying serverless applications so I've taken what Allan did and converted it to use SST and TypeScript.
Two lambdas are provided in this stack:
- delete-unused-log-groups - Removes and lambda log groups that remained when a lambda resource name changed
- set-log-retention - Sets all lambda function log groups to have a retention period defined in an environment variable
The lambdas are triggered on a schedule so you don't have to maintain anything as you continue to build. The scheduled jobs/crons are added only if the environment variables LOG_GROUP_RETENTION_CRON and
LOG_GROUP_DELETE_CRON are specified
It will set the log retention of all lambdas that do not match the configured retention period in the lambda environment variable.
Typically you'd want to set this to run every 7 days. As this is likely a much less frequently needed lambda function, you might be able to run it on demand. Resource names do not change frequently (really only when developers change it manually), so this lambda is set less frequently to run.
For both lambdas, it's possible to specify a regular expression to match the log group name. So you could use /-dev-/ to only apply this to log groups where their name matches -dev-
This is specified using the LOG_GROUP_MATCH_REGEX env var
See the docs over at SST for information on setting stage names and regions. The scripts in package json currently deploy to eu-west-2 (specified in sst.confiig.ts) with stage dev so change this as appropriate
This project uses the AWS v3 SDK to perform all interactions with the infrastructure. Version 3 of the SDK is not included by default in the lambda environment, so we bundle it with our lambdas
The lambda functions are running nodejs18.x and are executed on the arm64 architecture aka AWS Graviton2 processor.
You can show your support for the original project by following Allan on Twitter or connecting with him on LinkedIn. I'm on Twitter too if you have any questions on this implementation, or log an issue on Github.