SRM (Spectrum Report Manager) stops synchronizing the alarm table when a landscape is down or an Archive Manager is unresponsive
There will be a separate process/thread for each landscape. All these threads are synchronized and handled with a common job queue.
While processing records/alarms, if there is any error/issue occurred that attenuates the current handler job then it will halt the current process and also notifies the other jobs to stop processing records until it resolves the error/issue.
This is the existing behavior and it has been there for a long time. There are no architectural changes to this processing. This Handler job is common for other bucket handler jobs also (i.e NCM, SPM... etc).
Yes, If a landscape is down then it will wait until that landscape is up and reachable then it will start processing the rest of the buckets.