Server Problems

[thesillychicken Profile]

Comment on this thread to post on problems/bugs you are having with the server after July 4,2012

Please also comment on problems you had on/after 7/14/12 for a roll back error made by an unnamed mod *cough* t2t2 *cough*

[jerrcs Profile]

Almost all of the stuff in my inventory has disappeared, mainly a lot of gathered resources that I had prepped to build my new base with.

Figured I'd bring it up so it doesn't happen to anyone else, thanks.

[Goobler Profile]

Well i think the nether in survival is still having some major issues with building/destroying blocks and dynmap not showing new builds... other than that can't think of anything.

[Cheeetozz]

Almost all of the stuff in my inventory has disappeared, mainly a lot of gathered resources that I had prepped to build my new base with.

Figured I'd bring it up so it doesn't happen to anyone else, thanks.

All of my inventory disappeared. I was hoping this wouldn't happen to anyone else. That's a big deal.

[NgtFlyer Profile]

I didn't lose inventory aside from one iron block. Took one with my to Sky5 and lost my connection.
No big deal.

I believe Curley said that one of the drives had run out of disk space.

[thesillychicken Profile]

Almost all of the stuff in my inventory has disappeared, mainly a lot of gathered resources that I had prepped to build my new base with.

Figured I'd bring it up so it doesn't happen to anyone else, thanks.

All of my inventory disappeared. I was hoping this wouldn't happen to anyone else. That's a big deal.

I also had some stuff disappear out of chests.

[Curley]

ok guys, soooooooo yey!

most of you are aware, thanks to twitter, about what and why the server has had some issues over the holiday, but i figured it'd be helpful to post a concise explanation/write-up here.

So at some point during the morning of the fourth the twitcraft machine ran out off hard drive space. To be honest this was a complete oversight on my end, I was aware from the beginning that our space on the server was pretty limited and had always intended to clear out some unnecessary stuff, but we reached capacity way quicker than I anticipated, mainly due to dynmap and my decision to use the highest res maps (the dynmap folder is currently 35gb!) The system we have thanks to Dan Houtz at server central is running and amazing 4 x 73GB 15K Hdd's, but for performance they're in RAID10.

This issue decided to take place at probably the least opportune time, as these things tend to do, while I was out camping away from technology with a bunch of my friends. So as my twitter inbox started exploding I didn't have much i could do from my location. At that time I used my phone to ssh into the server and see what was up. People had been saying block updates weren't happening and people were lagging out. I saw from the console that the java errors, basically for every plugin, said it could not "write to file", which immediately got me suspecting hard drive space. All i could manage to do at that point via phone interface was to delete some unnecessary things and then reload the server and hope for the best. That as you have noticed and posted here, was not the case.

Thanks to t2t2, we managed to have a mostly functioning server for the past day or so minus inventory issues and whatnot. So now that i'm back at home and have had some time to look at the files I would like to explain what, I think, happened. When the server couldn't write the files it tried to sort of maintain the changes in memory so it could write them later, well, some of the plugins did that. Some others, not so much... I just spend the last couple hours comparing the current config files to the most recent back up and have noticed a ton of random whitespace throughout all of the plugins causing us trouble (PEX, Worldborder, Dynmap, etc.) not to mention minecraft jar's themselves. If you've ever worked with yaml files, you'll know how big o' deal that is.

Regardless of how it happened exactly here's the current status of the situation. First off the bad, Inventory data that was corrupted for players during the write issues decided to failsafe back to default blank inventories and can not be recovered. In addition the inventories were not being tracked or monitored by any plugin. Regarding this I will talk with the mods about what we can or cannot do. It very may well come down to a case by case honor system discussion with players involved or just having to suck it up :-/ please refer to my previous discussion on the podcast episode regarding minecraft not being programmed for long term play. Region chunk files that were corrupted (you'd notice these by seeing 16x16x256 areas that have reverted to the initial seed of the world) are in the same boat, mostly, as the inventory data. They, thanks to bukkit, have "fixed" themselves...basically by reverting to seed. The good thing is usually corrupted chunks only happen in the exact chunk a player was in and editing when the error occurred, To my knowledge there are maybe 3-5 of these that occurred. Now there is some experimental things we can try involving logblock. Since logblock writes the details of block changes to an SQL database it's pretty impervious to crashes. The trouble is the "logblock redo work" feature is experimental and may or may not work, again, we'll have to take that in a case by case basis as well. That is, if the affected areas haven't already been re-mined manually. Now for the good! I'm pretty confident at this point I've fixed all the configuration files, world border is back up, permissions have been reconfigured, and other small unnoticeable issues have been fixed. The only thing yet to be fixed is dynmap (that thing is...like...really messed up....though i'm confident i can fix it, just need to actually take the server down for an hour or so to do it.)

So to move forward. I'm planning some downtime late or early this weekend to take the time to fix up the ol' dynmap and get her functioning again. For now, from the users, I would love for you to continue to post here regarding any further weirdness or inventory issues you find in any of the worlds on the server. As for the discussed region issues please post here as well as /modreq at the affected region so I can have the chance to look at it personally. You guys have been amazingly patient and understanding throughout this time period and I appreciate it a lot. I <3 all of you.

[mdtaUK Profile Website]

I feel for you, and hope despite all the Tweetstorming, you had a good time.

Fortunately my epic mining sessions, and the building of my house have not been corrupted and all is as I left it.

[ErMurazor69 Profile]

Shouldn't take an hour or so to fix the maps, Just re-get a fresh copy of the mod and set it all up the way you want in a temp folder. shut the server down and delete the dynmap folder and copy the new one in and restart the server. Total down time should be minutes, assuming the folder deletes quickly (alternatively you could just move the folder somewhere else to delete later as this will be instant if its moved somewhere else on the same hard drive). After restarting the server just take each world and run the command "/dynmap fullrender <worldname>" one at a time (do not recompile more than one map at a time or you will lag) and it will completely rebuild each world map while the game is still running. I never noticed any lag at all while my server rebuilt the maps no matter how many times I had to do this. maps will also be updating and rebuilding while people are playing as well so parts of the map will appear early on maps not being re-rendered.

[t2t2]

Might as well stack this onto the same thread

Let's just say I learned the difference between /lb rollback since 9d and /lb rollback sel since 9d (spoiler: It is about 1,2 million blocks)

So there might be some funkyness with creations that happened since 9 days ago (especially chests and signshop signs, they need re-protecting and re-linking). If you find the funky, /modreq and we'll try to fix as much as possible.

But after that I'm taking a break from using lb... >.>