Solved Bungeecord loses connection to all servers except fallback

Discussion in 'BungeeCord Help' started by NIKNAIZ, May 28, 2017.

Thread Status:
Not open for further replies.
  1. Hey all again,

    Every night, when it's around 1 AM in my country, all players from across the network get teleported to the authentication server and a message in chat shows up, something along the lines of ReadTimeoutException: Null. It happens every single night at around 01:00 AM. None of the servers are crashed, they work on 20* TPS most of the day and the error given appears to be from bungeecord, so the issue might be lying in the bungeecord itself. To clear up, I haven't set up a system that would do such actions in a specific time exclusively, nor I have any plugins that would do such action. I suspect authme being the problem, but what's up?
    This is 100% a bungeecord problem because of the error given in the chat and no output being printed out in the server consoles itself. Here's an example of Bungeecord log exactly when this happens.

    So basically,

    [Player] <-> ServerConnector [Server] - read timed out is what happens at this time, but why? What's the cause? That happens on ALL of the servers. I can't pinpoint the issue.
  2. To have those plugins wouldn't hurt, but this is not a bot attack. It's been happening for 2 weeks, exactly at the same time minute by minute.
  3. I mean, we'd need more data to give you a better idea of what's up. How's server memory at the time? Do any crons run at that time that could overload memory? Is the network connection dying at that time? There could be a bunch of things wrong causing that, and that's just from a sysadmin perspective. Something like a Minecraft error would add a whole new line of possibilities. It's possibly worth monitoring more of what's changing at that point to cause the crash.

    I'm sure someone who's had the issue in the past will come along and assist, though.
  4. Maybe it's the witching hour ;).

    What bungeecord plugins do you use?
  5. BungeeAuthMeBridge,
  6. Network is fine, nothing is being cut. Memory is plenty. No crons are running. There are no errors in either of consoles.
  7. Can you link me to the BungeeAuthBridge plugin? Also, what commands do you have setup with BungeeCommands?
  8. BungeeCommands is a plugin that simply lets you create bungee-wide text-on-execute commands. Only text is being outputted on execution. This isn't the problem.
  9. What is that ram clearing thingy a few seconds before it all happens? Try disabling that ram clear thing for a night. Ram clearing plugins sometimes behave weird like ClearLagg deleting holograms.
  10. That was a thing I installed afterwards, the kicks have been happening multiple days earlier.
  11. Where do you host if I may ask? I think I have an idea why this could be happening.

    A server machine making a hugeee backup to a distant server - which eats all the bandwith.

    When I introduced backups as an add-on, every day at 6'o clock, there were random internet lag spikes because the backup process used the whole 1 Gbps to send all files to a distant server. I figured it out in a day and started limiting the backup bandwith. Things have been smooth ever since.
  12. SoYouStart in France. Machine is powerful, everything happens locally.
  13. I tried removing bunch of plugins, it still happens......

    Start looking from 00:00:00...

    In servers time, this happens exactly at midnight. I don't know what it is!

    Seemingly, the bungeecord loses connection to servers for a very short while at midnight. I have quad-checked and I don't have anything that does this. This is 100% bungeecord fault. Shortly after being kicked to the auth server, when I relog, I get sent to the hub server (session login) and I can't click on the NPCs as if the servers are down, but I'm able to do it after some while. I have updated all available packages on my Ubuntu 16.04 host.

    I have redownloaded the very latest bungeecord version, I have done everything and this still happens. I reconfigured the plugins, it still happens.


    Bungeecord thinks all of the servers are offline for a short while, thus teleports everyone to the auth (fallback server). Completely ZERO errors are in the server consoles, and there are no errors in Bungeecord log. This is as if they would legitimately switch the servers.

    I am ready to pay to whoever helps me solve this. It is very frustating and I do not have an idea as to what it could be.
  14. Just to make sure that it isn't the SoYouStarts' fault with the ping spike, I will create a 1536GB server for you for a few days. Use it to see if the problem persists. Details in PM in a moment.
  15. Sounds like a routine process that is either in the bungee configuration or
    maybe it's a routine process of your server host.

    Just tossing a few things out there.
    • Agree Agree x 2
  16. I thought about just that. Therefore I gave him a test server space. We'll see.
  17. I would think the same, but it didn't use to happen in the past and now it is happening, but no changes to files were made.
Thread Status:
Not open for further replies.