Periodic Memory-related crashes

[wisefool769]

I'm using this package to deploy pgbouncer to GCE. It mostly works well, but I'm observing periodic out of memory issues on the VM, which I suspect is due to a memory leak from pgbouncer (though I am not sure). Everything will work outwardly work fine, until at some point connections fail and you can't SSH onto the VM. My terraform configuration is quite vanilla, and I haven't seen anything informative in the logs. The only clue is the Disk Read IOPS steadily climbing, which seems to indicate intensive usage of swap because out of RAM.

1. Has anyone seen anything like this before?
2. Are there any suggested liveness or health checks that are easy to leverage when using this Terraform deployment?

Thank you!