Bill Moran <wmoran@xxxxxxxxxxxxxxxxxxxxxxx> writes: > I don't agree. I think that regular indexing is mandatory under some > workloads. Example: > ... > There are some additional indexes that I've snipped from the output that also > saw some benefit from reindexing, but let's just focus on file_fp_idx. Can you describe the usage pattern of that index? I'm curious why it doesn't maintain reasonably static size. How often is the underlying table vacuumed? regards, tom lane
