3

We all are familiar with the nasty errors that occur when the server loses what's stored for a particular area and replaces it (typically with something different as world corruption usually changes the seed). I've seen two friends now lose entire treasure rooms due to their chunk being regenerated.

Is there any way to prevent this from happening?

Improve this question
6
  • 2
    Um, chunk errors are the giant holes that only one player can see. Those fix when you log out and log back in. Nothing is lost.
    – John the Green
    Sep 8, 2011 at 20:44
  • 1
    I've seen chunk errors refer to both the disappearing invisible chunks and the "wait why is there suddenly a spot of mountain in my front lawn" kind of error. I'm referring to the second type as it's the one that makes your world look like a patchwork quilt after a few months, and I really don't have a better term for them. =/
    – Ken
    Sep 8, 2011 at 20:53
  • Hm, that would be world corruption. Typically caused by the server incorrectly shutting down (i.e. machine loses power). I'm afraid your only option is regular backups. And without a backup, once it's done, you're too late. :( Had that happen to my SMP world also. We had to restart. I make backups now.
    – John the Green
    Sep 8, 2011 at 21:05
  • Edit to say "world corruption" for searchability, perhaps?
    – SevenSidedDie
    Sep 9, 2011 at 0:09
  • 3
    I think by adding police into your world, it could prevent corruption. But those nasty pigs are most of the time corrupt too :S
    – Warface
    Sep 9, 2011 at 1:08

2 Answers 2

Reset to default
3

One approach to solving this problem is to run your world from a RAM disk, with regular (and atomic) pushes to disk. I implemented this on my SMP server (Linux guide here) for performance reasons, but as a side effect it should at the very least make world corruption much, much less likely to happen, especially using rsync to push the world to disk (rsync is about as atomic as you can get with file operations, so even a power loss during the push to disk shouldn't cause any problems).

With this approach implemented, a power loss will at worst cost you 15 minutes of work in your world (if you use the 15-minute scheduled push I do in the guide), which is a far cry less bothersome than world corruption (which has affected my SMP world twice before I implemented the RAM disk approach).

Improve this answer
3
  • Do you have any info about setting up RAM disks on windows server 2008?
    – Ken
    Sep 10, 2011 at 23:57
  • @Kort Some quick Googling suggests you'll have to download a utility that will create one for you; I know nothing about any of them, so I can't recommend any. Do be aware, though, that depending on your current RAM usage, using a RAM disk can actually slow down your system if you're not careful -- basically stick to a RAM disk that is smaller than your minimum free RAM during server usage.
    – Kromey
    Sep 12, 2011 at 17:36
  • I've been looking at developing a server cluster, so I'll likely end up just setting up a fresh linux server with a RAM disk according to the setup you described. Part of the reason is I run MySQL for big brother to rollback griefing and it uses 1.1gb of my current 4gb or RAM. =/
    – Ken
    Sep 13, 2011 at 14:53
2

I think the best way to go is to make regular backups, since this particular bug has yet to be corrected by Notch. Maybe it'll be fixed in 1.8!

Improve this answer
1
  • As far as I know, the next update will be on monday, so we'll find out soon.
    – Michael K
    Sep 9, 2011 at 15:32

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .