From Kirk:
We did all the work to bring up the single node RAC configuration this morning but after starting 3 application nodes successfully we hit a problem connecting the 4th server. The database went into high load (> 75) and the 4th node started getting DEADLOCK errors. Initial assumption is something to do with RAC memory or database connection configuration. We therefore fell back successfully to Caje (which had been running all along). Normal production environment is now back up. Patrick will investigate next week and design a test for looking into what can be done (with communications to whoever is around and reading email).