Silly question. I don't see any mention of buffer cache effects being worked around. Are you running a scan, tossing the results, then rerunning the scan?
I know with rsync on large directory trees (even when nothing needs changed), the second and later passes are substantially quicker and I would expect a similar, albeit less pronounced (~5K files vs. ~1M in my experiences), outcome here. _______________________________________________ Nmh-workers mailing list [email protected] https://lists.nongnu.org/mailman/listinfo/nmh-workers
