syncthing falls down when you hit millions of files.

Platform agnostic? Rsync from the perspective of duplicating the client-presented data.

Or rclone, another great tool. I tend to use this for storage providers instead of between self hosted systems (or data center fully self-managed systems.)

If the NAS uses zfs then zfs send/recv is the best, because you can send only the changed blocks. Want to redo the names on every single movie? No problem! I do recommend sanoid or syncoid, I don’t remember which. ZFS snapshots are not intuitive, make sure to do LOTS of testing with data you don’t care about, and which is small.

In terms of truly duplicating the entire NAS, we can’t help without knowing which NAS you’re using. Or since this is selfhosted, which software you used to build a NAS.

That is… not what I said at all.

Solar panels still provide power on cloudy days.

In fact under certain conditions they’ll produce more power than under full sun - solar panels drop in efficiency when they are too hot. (Yes I know this isn’t normal; normally full sun will produce more power, but some people don’t know cloudy days are fine for solar energy production.)

Author didn’t seem to have a clue. Many of us didn’t protest or leave because of the fact that they implemented charges for their API - nope, was totally open to that! - it was the way they started charging.

I don’t think I’m alone either here. So many were open to paying fair prices for usage. But reddit repeatedly promised it’d be fair and reasonable. For months. And then when they finally dropped pricing info it was outlandish and would be taking effect before third parties had a chance to make appropriate changes.

This amounted to a power play meant to drive mobile users back to the reddit app. Why? Money and control. Bad for mods, users, and developers, it was a selfish play I will never forgive them for.

How did the author not know this, or if they did, why was it not front and center? Feels like they were parroting company talking points.

A battery that is utter trash for driving purposes still has tons of life left for other uses.

Have you isolated source formats? Does it do this with h264 and hevc?

Tried stopping other containers and/or all other services? Could have less to do with pure power and more with sheer volume (scheduling.)

Any issues if you use the current server as a file server and play the video on another machine? Any modifications to playback speed?

If none of that works you might try iperf3 to check network speeds. Or fio for disk speeds. Run tests covering all kinds of situations, you’re not so much looking for max speeds but instead for inconsistencies.

Don’t get too lost in the small stuff. Isolate systems. Might even move or clone the boot drive to another machine as a test. Try a different switch. That kind of thing.

Good luck!

Doesn’t the article indicate the difference had to do with utilization of e-cores? Since the newer Apple silicon has more e-cores and fewer p-cores?