I have read the other two suggestions on providing more killOp() and read Andrew's comment on the difficulties in the medium term.
I just watched a situation where a primary became unresponsive and the queues piled up. The solution was to cause an election, but the dba wanted to kill operations by user (the application user) and couldn't.
It would be nice to have something more than the real-time panel, which in this situation had become unresponsive as well so it was no use, such as a DBA console where an authorized dba could kill operations started by other users.
Please see the new killOpSession privilege https://docs.atlas.mongodb.com/security-add-mongodb-users/#std-label-atlas-specific-privileges for a purely data plan option here
It would be great to have an option/button to "Kill all operations like this".
We too experienced a case where many slow operations caused the database and the admin panel to become unresponsive. A big portion of the downtime was linked to the inability to "Kill all slow operations".
Please provide an option (button, filter, list, or something else) to list/kill slow running operations.