Provide more control of how runbooks are allocated to a sandbox
There are several scenarios that cause sandboxes to crash or runbooks to fail in both hybrid workers or in Azure.
1) Sandboxes fail with out of memory randomly. If the wrong combination of runbooks get executed in a sandbox the the sandbox can crash intermittently. If there was a way to control how many jobs run in a sandbox, or flag particular runbox to run by themselves and not share a runbook this problem could be mitigated.
2) If runbooks land in a sandbox that require different subscriptions one of the runbooks will fail. Someway of preventing this so that either the sandbox could handle multiple subscriptions or that runbooks could be grouped to be defined to run only in the group could mitigate this.
Just guessing :
My first query would be are you running the Runbook using Azure or Hybrid worker?