When a batch job triggers many sandbox plug-in executions almost simultaneously (in my case, a monthly job firing the rollup-calculation plug-in for 63 Debit Note lines at once), one record intermittently fails with: "The plug-in execution failed because no Sandbox Hosts are currently available." (SandboxFabricHostCommunicationException – WorkerShutdown).

This was confirmed as expected behavior of the shared Dataverse sandbox platform, under high concurrent load, the platform occasionally cannot allocate a sandbox worker for one execution, causing a transient failure. The rate is very low with no data integrity impact, but it still forces customers to manually retry the affected record.

I believe many people face the same problem. Therefore, I suggest implementing a function to address this Thank you!