System integration is about exposing services or consuming services, either from local plugins/microservices or from external servers (including database, message bus, RPC calls).
Please follow:
Timeouts are very important when implementing system integration.
TODO link to code of db or messaging that allows to configure global timeout
TODO link to code of db or messaging that allows to configure method level timeout using varargs mehotd(args, WithTimout())
After a successful recovery from an HA event (restart, loss of connectivity), a client MUST reconnect to a service it was consuming before the event and perform data resynchronization.
TODO link to code/doc of db or messaging
If a plugin implements a client that consumes an external service, and the client is unable to connect to hte service before a timeout, the plugin MUST propagate errors. The application will not start, and an external orchestrator must clean it up and start a new instance..
TODO assuming that there is a default deployment strategy for container-based cloud (as is with K8s) that will try to heal the container and basically recreates it.
TODO link to code of db or messaging plugin that propagates error