Event  | Info    | SCALINGREPORT   | Scaling execution enabled, begin scaling vertices:
{ Vertex ID 5591cebca2a45b74833df01196d1a431 | Parallelism 74 -> 80 | Processing capacity 6671.45 -> 7131.00 | Target data rate 3889.81}
{ Vertex ID 162a29c968c94eac6fbdfc2f8ccc2080 | Parallelism 180 -> 128 | Processing capacity Infinity -> Infinity | Target data rate 50.67}
{ Vertex ID 009b60dbc7737a8a2e91e7b3c30d9949 | Parallelism 128 -> 169 | Processing capacity 70.68 -> 93.00 | Target data rate 50.67}

these autoscaling decisions don't make a lot of sense to me. why on earth is it scaling UP

009b

? here's the output log from right before:

ue1d.FlinkDeployment.AutoScaler.jobVertexID.009b60dbc7737a8a2e91e7b3c30d9949.TRUE_PROCESSING_RATE.Average: 71.325
ue1d.FlinkDeployment.AutoScaler.jobVertexID.009b60dbc7737a8a2e91e7b3c30d9949.SCALE_DOWN_RATE_THRESHOLD.Current: 176.0
ue1d.FlinkDeployment.AutoScaler.jobVertexID.009b60dbc7737a8a2e91e7b3c30d9949.SCALE_UP_RATE_THRESHOLD.Current: 56.0

is this a catchup buffer problem? maybe it thinks it would be unable to catch up within the desired window?