We have a lambda defined in multiple regions running in active/standby mode. Switching requires a manual Route53 update.
The Lambda basically just pushes a message onto an SNS topic.
Looking at switching to an active/active configuration. But trying to figure out how to set up a healthcheck.
In the examples I've found online, the /healthcheck lambda code is simply returning a status as set in an environment variable, for ease of testing. I suppose that makes sense, but is there anything else the I should consider? For instance, if the lambda was talking to DDB, a deep healthcheck would confirm that DDB is healthy. But is there anything I can do to check SNS other than actually publishing some dummy message of some sort?
Furthermore, I'm a bit confused about differences between
- pointing the ALB to the healthcheck, and having the R53 Alias records doing a
Evaluate Target Health - Having the healthcheck endpoint specified on the R53 CNAME records.
Is there some difference? Should I do one or the other? Or both?