Current implementation of Operations server will reply with "success" response as soon as logs are accepted by operation server.
We should introduce log delivery status for each log upload request in order to guarantee log delivery to external system.
Log delivery status will have optional (in case of log delivery failure) error code. Possible error codes:
NO_APPENDERS_CONFIGURED - There are no appenders configured to process log events.
APPENDER_INTERNAL_ERROR - Internal error in appender code while processing log events.
REMOTE_CONNECTION_ERROR - The error in connection to log delivery destination system.
REMOTE_INTERNAL_ERROR - The internal error of log delivery destination system.
It is also possible that endpoint SDK will detect timeout of log delivery due to no log delivery message from operations server.
Endpoint sdk should provide default and allow to specify custom strategies of recovery from log delivery failure.
We should develop low level design and documentation for this feature.