A customer is experiencing OutOfMemory exceptions caused by the 'Update Allocations from Estimates' job. When the job does actually complete, it's taking approximately 2800 minutes (48 hours) to run. It is scheduled to run once a day at 6 pm, so it essentially never stops running. When it does execute, the job thread executing it consumes nearly 100% CPU for the entire duration of the run.
17k active projects
1k active resources
Time slice durations appear normal
It would appear that this job is not efficient and thread dumps are indicating that this job is also performing insta-slicing when it runs. We should expect insta-slice activity only to be performed for UI operations, not for Jobs or during XOG.