The 0.12 upgrade instructions say to mark all existing jobs as requiring a status update (see [1]), and that the only consequence is a slightly higher than normal load on the system until the backlog of status updates is cleared.

It turns out this is incorrect. Instead, due to the way the status update system interacts with the rest of the scheduler, it results in a near total outage of the automated job scheduling:

1. Any currently running jobs will not return their systems to the pool until the backlog is cleared
2. New jobs will not be scheduled until the backlog is cleared

The upgrade instructions should be updated to:

1. Clearly state that the Beaker installation will be in outage until the update is complete
2. Describe an alternative upgrade approach for use with large Beaker installations which allows previously executed jobs to be checked for status discrepancies *without* requiring an extended system outage

[1] http://beaker-project.org/docs/whats-new/upgrade-0.12.html#asynchronous-job-status-updates-bug-807237