So as quite a few people are aware, there is a tps issue on the server. The only real way to fix said issue is by restarting, but that brings a problem itself: only admins can restart the server. When I went online today, I realized the tps was at a solid 1.5. I then went through the discord chat and realized it had been under 5 tps with only 3-5 people online for at least 2 hours. Naturally, this is a problem that turns the server into an unplayable lag-fest with little solution, so I have some ideas. Install a plugin that allows for online players to vote for a server restart, needing either popular or unanimous vote. This will make the players less dependent on the admins, and allows for quicker responses to low tps. Give moderators the ability to restart specifically MC Eternal. This is a more complicated solution due to permissions, but I'm sure there's a way to configure the permission plugin to allow for them to restart the server. This will also make it easier for players to find someone who is online and able to help. Have the server restart more frequently. I understand the increased downtime isn't a huge appeal, but this is the simplest solution that will still have a positive impact on players' experience.
You summed up the issue pretty well. I returned to infinity from eternal simply because of the lack of stability tps-wise. Now, I could play on node 1, but the community and playing with another is 90% of the fun! I can only make speculation pertaining to the cause of the lag. Common reasons for lag: 1) - Too many mods/plugins. 2) - Server is running out of RAM. 3) - Too far away from the server. 4) - To many entities/items in the world. Above are a couple common reasons for server lag on the client. We can eliminate the third reason because I was able to play on the server without any lag and I live ten thousand kilometers away. That leaves 1, 2, and 4. The problems on the eternal server are related to memory leaks, which could be associated with a certain item, mod, plugin, block, or player. We have had specific players that contributed to the lag, but an average of 3-5 players and the lag continuing likely means the lag is due to something else. The only surefire way to find out the cause of the lag is to manually look through a memory dump - an admin task - that takes a lot of time. While we are looking for a solution, a method of prompting more server restarts is definitely needed. I agree with your suggestions for a lower server-restart timer (1-2 hour) and a vote option. The problem I see with the vote option is that there is always that one dissenter, if not, an afk person that won't vote.
As Aido said in another thread the best thing you can do to help us locate the issue is get us the player list while its lagging. Then we can look through them all and see which players are always online when the lag. Problem is with mem leaks though is they start early and tps doesn't tank until later. So, we need to know whos online when tps first starts to drop.
Would a logger plugin that reads the TPS and memory state and notes the players online when one of these two values are too high/low for too long help at all? If the names are important for when this happens, that would be my first idea, but I know development takes time and you're lacking on devs right now (cough, thinking of sample ideas, cough).
A plugin that does that actually wouldn't be that bad of an idea. As long as it doesn't read it too frequently, then I don't think the plugin would cause further harm either. This also makes it possible to graph tps per person/time if need be.
I figure running it something like every 10 to 20 seconds would be a low enough rate that it doesn't cause the server any pain, but does the job. The main thing would be to make sure it's not spitting out huge logs because its recording over and over if the TPS gets low and stays there. Either a cooldown or a do not log switch would probably work. Basically, the idea of the latter would be after it logs once, it doesn't log again until the TPS comes back up and then goes down again.
So a few points id like to make. When tps is this low. Why is it not reported? Mentioning it only in ingame chat wont help us nor you. Let us know in support and we can help. A lot of people don't understand we don't 24/7 monitor the servers. For point 1. Thats a maybe however the plugin would have to be made for sponge. Not spigot. I don't remember seeing one that could do that on the sponge plugin directory. For point 2. That's impossible and would likely never get approved. Even Sr.Mods cant reset servers due to in-game command limitation. Max we could run is /stop which doesn't reactivate the server. Simply kills it. Point 3. That's a valid suggestion and something that we could do. Restart time for eternal is I believe either 6 or 8 hours currently. Maybe lowering to every 4-5? That's about all i have for opinions or concerns
If the server is being automatically restarted or restarted by players/mods we won't be able to further investigate the issue and potentially resolve it as there is currently no known way to reproduce the issue, which would allow us to troubleshoot the issue on a test server.
The issue doesn't just start the moment the server is started up. Something is being loaded or triggered which starts the issue. Without knowing that we can't reproduce it on a test server or even empty live server.
One thing the players can do is for someone to take a screenshot of the scoreboard when it lags up (only when the GC30 is high on Status page) and post it to this thread (or a new one). Hopefully, we'll notice a common player to every lag event and know who's base/activity to check into. That will narrow down the area/players at least and help a lot. The real problem here is, since the lag comes much later as memory leak builds up to insane levels, it could be a player that has already logged off. Or it could have been a move between dimension someone made 45 minutes ago. It could have been anything that happened in the last hour. This would also be a problem if we made a plugin. It may not actually log the player/event causing it since it most likely happened several minutes ago. If you can get us screenshots of the scoreboard when it starts to lag up that might help narrow it down though.
i will say in having what was a massive base that having less Little Titles and Architecture with a combo of blocks that have not a lot of massive amounts of NBT on them even if it is 3 of them and you use 3,788 that adds up very fast and can make its own lag i mean the bigger ME drives get bigger NBT on them the more that is in them and that brings me to your core of the ME that you need to make sure the controller and the drives are all in one chunk to help reduce the lag of the AE2 itself
I mean there is an entire guide on proper ME design and usage that I keep directing people towards, but most people refuse to look at it or just say things like "well I'm sure it wouldn't help in *my* case, so Ill just try something else instead"
We also have the suspicion that multiple bases with huge chunk sizes are getting loaded at the same time causing the server to use a unusual amount of memory. Having timestamps and screenshots of the online players during the time the GC time on our server status turns red would be helpful.
Applied Energistics — MyM-Wiki 0.0.1 documentation Its fairly basic, and the fact that you said you have a separate Thaumcraft ME makes it sound like youre definitely further along the right track than most people I direct to it
Just to get started with this, lag spikes have been happening now for a few minutes and the GC has just gone dramatically up. Spoiler: GC30 Here are the people currently online. Spoiler: Online Players And here are the last few people to join from just before things started going south Spoiler: Players Joined
