Spigot Regionerator 2.3.0

If you randomly sample chunks, does visit status look correct? I should include a visit expiration timestamp with that command, that would help a bit. Unfortunately I don't have much else assistance to offer, the debug is pretty ineffective in this case because the number of chunks being checked makes it unrealistic to expect to sift through logs.

That's pretty tiny, honestly. I started writing Regionerator when I was doing backend work for a server whose backups were pushing 30GB, Regionerator pulled them back to under 15GB.

If you're worried about using Regionerator, you can always delete region files based on last modification date in your start script. It won't respect claims and it won't get individual chunks, but it'll be consistent. Theoretically you could actually script in dropping individual chunks, but the simple way to do that doesn't save disk space until the region containing the chunk is loaded, so it doesn't really matter much.

Edit: For example, a simple script to delete regions modified more than 30 days ago:

Code (Bash):

#!/bin/bash
regionFolders=("./world/region/" "./world_nether/DIM-1/region/" "./world_the_end/DIM1/")

for regionFolder in "${regionFolders[@]}" ; do
    find $regionFolder* -mtime +30 -exec rm {} \;
done

I'm sure it could be made more compact, I'm far from a bash expert, but there're plenty of "safe" ways to do the dangerous work.

 

Sorry to have left you sitting for a bit, I'm surprisingly swamped despite shutdowns.

/regionerator check shows the visit status, but unfortunately in its current state the info it provides isn't super useful for debugging this particular issue. If it were an issue with any other aspect I'd say leave the plugin flagging but tweak the visit duration absurdly high so it won't run deletion, but that won't be any use if the flags are inaccurate. It does sound like visit data is not being saved, which really should only be possible in the event of an improper shutdown.

As far as things I want to add to improve on this:

• logging to a separate file for saving errors
• saving queue info command
• actual numbers, not just visit status, visible in status check

Edit: Had an idea and found an error, should be able to push that through tomorrow. Gotta pass out now unfortunately, more fun scheduled for tomorrow.

 

Boy was I wrong. The initial fix was simple, but without the additional info there'd be no way to double check if it was working until deletion occurred. I've had to revisit a lot of areas of the design, it was originally built to just be a much less messy version of what was already in place.

 

I've pushed Beta 4 to GH if you're in urgent need. The biggest highlight is that I finally introduced checking the chunk's last modification timestamp as an additional failsafe for visitation flags (in addition to flags saving more reliably). Biggest drawback, I apparently screwed up somewhere so now cycles just run concurrently all the time. Whoops.

I'm going to make myself an issue listing all the features I still want in the full release and flaws I found testing and then pass out, hopefully will be able to polish something up soon even though I doubt I'll manage a full-featured release in the coming few days.

 

I have a pretty solid life balance these days, don't worry. If I'm honest, the reason I pushed out beta 4 so late last night was that I was actually dragged into games with friends in the middle. I was pushing more of my free time into this because issues with areas being deleted could actually be my fault instead of players just not being as active as they thought they were. The plugin may be a little hobby project, but I've got my pride. Beta 4 is at least stable, though I do have more polish already completed at this point.

 

Jikoo updated Regionerator with a new update entry:

Combined Beta 4&5

Read the rest of this update entry...

 

Don't know, I did at one point stumble on it by mistake but I haven't looked. If you can't find it with a quick search feel free to repost.