Posts by Telesphoreo

Why on the BlockPlaceEvent can't we detect if it's a spawn egg? If so, just cancel the event and then spawn a new clean mob with no NBT data in its place. Perhaps TFM could scan the egg and if it does contain a name like Dinnerbone, it will be applied but strip anything other than A-Z and have a max of 16 charachters

↩  Miasmus

↩ @'Ryan' I'll think of a comeback later. It's on my low priority list for right now. Don't mind if I bump this in 8-10 months, thanks

So I have a backup of the old TF Wiki from around September 2018. However, I made a backup of my entire Linux install (except for /dev and a few other directories I think).

I have a copy of the web files for the wiki and the database. However, the database was in the /var/lib/mysql folder. I tried to shut down MySQL and upload the files in there and start it again. It almost worked. In phpMyAdmin, it says "Table does not exist in the engine." I found a bunch of people who had solutions but were trying to copy everything from an old server over and were purging a bunch of stuff. I'm wondering if its possible to salvage the files and create a .sql or get it working again somehow. Keep in mind nothing is actually corrupted, and I have the entire Linux install to look for files if needed. Any advice? I also thing it might have to do with different MySQL versions. I don't remember what I was using back in September 2018, but I am now using MariaDB 10.6.2

Bump. This is important, can we get any update on this? I'd like to access the ProBoards forums.

↩  Luke you're not really entitled to an opinion Mr. Total Freedom

↩ @'Ryan' I'm sure you could just hire more devs with forum accounts that are one day old! Seems pretty foolproof and nothing could go wrong

Get down on your knees and beg for it.

↩  lyicx i can send you a DLL that bypasses it. You can use the setup.exe file while booted into windows and do a clean install by selecting it from the options

The Microsoft event is today so I'll be waiting mainly to hear about the enterprise side of things

↩  volleo6144 that worked for the problem, thanks. I'll see if it works on the second one it's asking which is just to a higher exponent.

edit: yep this worked thanks so much for the help

↩  Ivan no it's a Texas Instruments TI-84 Plus CE-T

Right so I'll make this to the point. The problem is this:
f(x) = x^4 - 12x^3 + 38x^2 + 116x + 65

I have to find all zeros for the equation. Now, the way we were instructed to solve it was to go on our calculator and put it into the graphing calculator and look at the table

As you can see there is only one zero on the table. What our teacher said to do was notice how there would be more zeros if it switched between a negative to a positive. There has to be a zero between there. However, this is not the case. We learned about diving the products over the quotients, but even so there is only one real zero. I even ran it through a calculator and the only real zero is -1. That would be fine but we need to find nonreal ones.

I used synthetic division with the only zero (-1) and after doing synthetic division you get this: x^3 - 13x^2 + 51x + 65. To find all the zeros you'd just do synthetic division again with another zero, but there is no other zero.

I never learned this in math class. What the hell am I supposed to do if there is only one zero? I can obviously use the zeros calculator to find nonreal numbers. What the calculator says is 7 - 4i and 7 + 4i, but how do I get there??

Thanks in advance to @"StevenNL2000"#2

After doing more testing here's what I found:

1. Using OpenJ9 does improve RAM usage, significantly
2. Running it from the command line would be second best option of OpenJ9 is not available (which is weird, I'm pretty sure the Ubuntu OpenJDK packages are HotSpot
3. Using the HotSpot Docker image. Just no, don't do this.

Using OpenJ9 Docker images and running from the command line used nearly identical amounts of RAM on the server. On the server itself, using the Docker image took slightly more memory, but nothing of significance.

I may have gotten a bad Paper build last night. Paper is back doing its magic. A server I run for a friend which only has AuthMe and CoreProtect as plugins was idling at 2GB of RAM on Spigot. It's now idling at 1.4GB on Paper. The CPU usage is also reduced on startup with Paper. However, when idling Paper and Spigot's CPU usage is almost the same.

As for StevenNL2000Freedom, it's pretty much the same story. The only difference with Paper vs Spigot is LibsDisguises throws an error on startup with Paper.

This is what RAM and CPU usage is looking like with Paper on OpenJ9.

(Note that the Paper support team was yelling at me for using OpenJ9. They wanted me to use HotSpot and said all the other Docker images were bad, which is ironic)

Looks like things are a lot better now

↩  StevenNL2000 I think the problem was actually AsyncWorldEdit and BlocksHub, the two best plugins we have that are never the roadblocks to a new version of Minecraft! Makes sense since AsyncWorldEdit does a lot of patching and other stuff

The server continues to idle at nearly 8GB of RAM

Holy shit

Edit: So the problem is using HotSpot vs OpenJ9. The server literally doesn't start on OpenJ9. At first it said it was BukkitTelnet. I removed it and then it started up. Now it's magically stopping on JDA.

Fuck you Mojang

↩  StevenNL2000 Well obviously I'm not using it for adult content LOL. It has a 1 gig connection and I'm not uploading data to it 24/7 so I think it'll be fine.

The main difference I'm noticing is from Haswell CPUs on the OVH one to a Xeon Skylake one. Of course its all shared but its a newer CPU.

I could've doubled the RAM from OVH but then it's only $10 a month more for a dedicated server from Hetzner.

Mojang really screwed everyone over by changing it to an unstable Java version, especially toward the end of development of the update

↩  lyicx no kidding. and I'm still shocked Mojang decided to go with Java 16 with no other option. They could've added a warning in the half assed update and actually changed it in 1.18. And I would've made it Java 11 instead since Paper put a big warning saying Mojang will switch to Java 11 a while ago.

A server with no plugins at all is still idling at around 3gb of RAM used. It's genuinely insane how unoptimized this new update is. At first I thought there was a memory leak in ProtocolLib but no, it made hardly a difference when it was removed. The CPU on the hetzner cloud appears to be much better than OVH's was so that's a plus

Also side note, yes I am using the Aikar flags. I'll try removing them. Perhaps there's some behavior changes in Java 16 that the flags don't account for

Regardless, I'm moving to the new server as it's a better deal and better specs

Semi serious: 1.17 is genuinely so unoptimized. I know this is a joke server but I actually can't run two servers at once anymore with 8GB of RAM on 1.17. The server will be down for a few days. I found a better deal on Hetzner for their cloud servers. It's ~$19 a month and has double the RAM. I'm gonna move to that as it's a better deal than what I was paying for from OVH anyway. Sorry, even running one server was maxing out RAM. I tried everything. Switching back to Spigot helped and running it from the command line without Docker being involved made things worse actually (sorry Wild)

↩ @'Ryan' Do I really need to file a bug report?... Seth wrote it when Pterodactyl was in beta and we had no choice except to write it in the beta API. I rewrote it using Pterodactyl4J rather than raw JSON because it now uses weird headers and the documentation for 0.7 was much better. The bug was that it was using the old API and it flat out doesn't work on new versions of Pterodactyl

Edit: made a PR on JDA and TFM with their respective fixes / additions

Slight but relevant bump. I just tested the Pterodactyl integration and as it stands in the AMG repo, it is broken. I rewrote it entirely primarily because I am using it. Either it should be rewritten or removed. It seems useless to keep it in if it's actively broken