taulover

It's difficult to say, because based on Hermitcraft video examples, although the lag that is manifested is server-wide, it also seems often exhibited client-side by what appears to be a drop in the frame rate.

I attempted to do some server-side tests by running ilmango's double 2x2 spruce tree farm in both 1.16.5 and 1.17. I picked this farm because it uses a 2x2 tree and also because nearly all of its redstone runs regardless of whether the tree grows or not, allowing us to isolate those effects. I ran both with only Carpet mod installed for performance gathering.

I used Carpet's server-side autoclicking (/player <playername> use continuous) in conjunction with a three-minute tick warp (/tick warp 3600).

In 1.16.5 I got the following numbers:

Farm on but bonemeal circuit off and no saplings placed: 1.52 mspt
Farm on but no saplings placed: 1.63 mspt
Farm on and saplings placed: 2.46 mspt

In 1.17 I got the following numbers:

Farm on but bonemeal circuit off and no saplings placed: 1.74 mspt
Farm on but no saplings placed: 2.13 mspt
Farm on and saplings placed: 3.40 mspt

In 1.16 there is a 50.9% increase in server lag for growing trees, whereas for 1.17 it is 59.6%. If we factor in both bonemealing and growing, it is a 61.8% increase in 1.16.5 compared to a 95.4% increase in 1.17. For a server under high load, this might be a significant enough difference to cause noticeable server lag.

I also tested using ilmango's fast 4-type tree farm. Using the same methods as earlier to run the farm, I got 6.95 mspt in 1.16.5 vs 7.26 mspt in 1.17. Unfortunately, most of the redstone in this farm runs in response to a tree growing, so I couldn't isolate the effects of different factors.

However, the simpler setup of running this farm meant that I could try to simulate a situation of entirely server-side interaction. I spawned a Carpet bot to place saplings and teleported to 1000, 1000 before running the tick warp. I didn't experience any lag; the server reached 11.00 mspt in 1.16.5 and 10.35 mspt in 1.17. It is certainly possible that for a server running at higher load, this may lead to server-side lag, though if so this wouldn't be new since 1.17 seems consistently more performant in this area. However, considering that Carpet bots are entirely server-side, this may not be an accurate simulation; it's certainly possible that something in the communication between client and server that is connected to the known client-side lag may lead to further server-side lag when it comes to this tree growing thing (which might help explain why when a player is present, the server does lag more in 1.17). And that combined with a server at high load may cause issues. But that is just speculation.

Anecdotally, 1.17 just felt laggier in general, especially client side (freezing for far longer and/or crashing despite the relatively low mspt values, especially when saplings were grown). However, it must be stressed that these results are very inconclusive and I don't know if they're reproducible.

I conclude that there is likely an increase in both client and server side lag for both bonemealing and sapling growth in 1.17, and speculate that the two may likely be connected. But this is not conclusive and warrants further investigation.

This doesn't seem like a purely client-side issue; the lag spike occurs globally for all players across a server.

Furthermore, the lag spike appears present even when the tree grows naturally without bonemeal.

Are they actually duplicates? MC-90307 is caused by air having a deceleration of zero, which makes the boat travel infinitely horizontally. On the other hand, this bug has the boat start on a slime block, and leads to a launch into the air, seemingly prompted by nothing.

Granted, I am not very experienced with Minecraft mechanics, but I'm surprised that this is marked a duplicate.