MID Servers as ephemeral Infrastructure as Code in AWS

[stevejarman]

We're in the process of implementing MID Servers as "Infrastructure as Code" builds within AWS - i.e. using Silent Installer options, inserting Mutual Auth and Secrets Management certs via command line, etc. Essentially, making it so that MID Servers can be destroyed and rebuilt on the fly as a healing mechanism.

The one stumbling block I can think of is the ecc_agent record itself, which will be left behind when a MID is "destroyed" and potentially cause a warning/error when the new MID comes in reporting itself with the same Name.

We're going to experiment with this to test exact behaviours, but I wanted to start a thread in the meantime to see if anyone else has experience dealing with this/thoughts on ways to deal with it.

My initial thought to a "solution" was to implement a Before Business Rule on ecc_agent which will check anytime a new record is about to be inserted for an existing record with the same Name, and - if found - delete it... Again, completely untested at this stage - just thoughts.

[Kieran Anson]

Hey,

Not sure if it helps, but there is a scheduled job that will remove orphaned containerised MID servers after 30 days

"Purge Orphaned Containerized MID Servers"

[stevejarman]

Thanks! Not sure if that will apply here, but I'll look into it / it's good to know regardless.