Server Technical Issues Tonight

Please Note: The TotalFreedom Forum has now been put into a read-only mode. Total Freedom has now closed down and will not be returning in any way, shape or form. It has been a pleasure to lead this community and I wish you all the best for your futures.
  • A very quick post just to explain why the server was crashing a lot tonight, as it was something that I was quickly able to spot and fix, and is something that I had not expected, and thought the monitoring should have picked up but did not.

    In short, the /home partition on the CentOS host had reached 100% capacity. The server runs in /home and at some point tonight the /home partition hit 100% usage.

    The obvious question is why didn't out monitoring pick it up, after all I've made a bit of a song and dance about it being so good. And in short it's because the server still had a lot of disk space available, however it was on a different partition, so we had 50GB "Free" but not that the MC Server could write to.

    What I've done for now is delete a few old files (Well, download them and delete them from the server) and this has included the hub world map, which as far as I'm aware was not being used at all, so that world will now be totally blank.

    Longer term one option is to simply expand the partition, but to be honest that's not something I'd like to do if I can avoid it, as it can cause problems, and is realistically going to be a few hours outage to back everything up before starting, but might be what's required. I'm hoping I've bought us a little bit of time before this becomes a major issue again.

    In terms of world sizes, this is the current stats:

    Flatlands - 29.8GB
    HubWorld (Pre-Downloaded) - 1.55GB
    MasterBuilderWorld - 4.13GB
    PlotWorld - 18.5GB
    StaffWorld - 3.3GB
    World - 19.6GB
    World_nether - 5.33GB
    world_the_end - 4.71GB

    I don't want to be wiping worlds at the moment unless there is no other choice, especially as I know when I suggested having a rolling main world people weren't so keen on it, but it's just to give everyone an idea on the current scale and why this is such an issue as we continue to grow.

    I'm hoping in the longer term some of these worlds will be split out into their own servers, which will help with the issue but it is just something to be aware of.

    Apologies again for the instability tonight, I'm hoping I've bought us a few days at least to start properly sorting this, but I will keep a close eye on things over the coming days.

    Wild1145

    Network Owner at TotalFreedom

    Managing Director at ATLAS Media Group Ltd.

    Founder & Owner at MastodonApp.UK

  • Quote

    @wild1145#2261 And in short it's because the server still had a lot of disk space available, however it was on a different partition, so we had 50GB "Free" but not that the MC Server could write to.

    ahh, i see, iirc back before the wipe, we had a full like 100gbs of storage, if not more, for the server, bit overkill, but its always good to have some overhead

    also i am curious, how much of the flatlands' huge amount of data is from coreprotect

  • @DragonSlayer2189#2266 yeah I had forgot Linux partitions itself this way. I assumed 150gb disk would be enough but didn't account for it to partition chunks of it away from the home mount.

    And none of that data is for anything other than the world files themselves.

    Core protect currently for everything is taking up around 24gb since I took over hosting the server.

    Wild1145

    Network Owner at TotalFreedom

    Managing Director at ATLAS Media Group Ltd.

    Founder & Owner at MastodonApp.UK

  • @wild1145#2270 well for one, mabye we should try purging any coreprotect data that is over a week old, to help midigate the amount of storage that is using.
    as for worlds, I am not 100% sure how exactly we could implement this, but if posible, maybe there should be a system in place to automatically delete chunks that haven't been modified in a while, or that a player was only in for a few seconds, like if someone did /tpr, and then loaded the chunks there, and then imediately went back to like spawn, then it would automatically remove those unneeded chunks.

  • @DragonSlayer2189#2274 the core protect data is less of a concern tbh. And I'd rather only be deleting data when we wipe those worlds or parts of the world's.

    As for the maps the answer is to purge the maps properly and more often...

    Longer term I've got some solutions as I'd like to be splitting things like the plot world and staff worlds to be their own distinct minecraft servers but thats a little way off. Shorter term I will need to see how quickly the server files are growing again and see if I need to expand the disk or if I can wait it out.

    Wild1145

    Network Owner at TotalFreedom

    Managing Director at ATLAS Media Group Ltd.

    Founder & Owner at MastodonApp.UK