Minetest Forums

Hello, I am looking to start a server on a neat serviced VPS that I found (not cheap but gets the job well done once any mods problems are sorted out). However I have some questions regarding the Minetest server...

1. How are you supposed to back these up so that there no data corruption problems while it running?

2. Do I need to deploy other measures other than protection, anti cheat enabled, and having xban2? To keep things from going "side way"?

3. How is "lag" measured in this game? I noticed that there on the "fresh" server a lag metric sitting at around 0.1~ when I joins but I have no idea how much (if any) lag that means? If that not lag then how "much" of this measure is "too much" to start being considered lag?

If there anything else I may need to know feel free to add and thanks in advance!

Backups:

Sqlite3 has a simple method to create a backup safely while your server is running, simply issue the following command:

sqlite3 map.sqlite ".timeout 1000" ".backup map-backup.sqlite"

Do this periodically and make sure to store copies off the server. Other files on the server may need backing up too, of course.

You might want to look into the City Block and Liquid Guard mods as a safeguard against water and lava griefing. Also the Chat Anticurse mod will prove useful for basic chat moderation.

Filter is better than chat anticurse - no false positive, support for warnings and mutings rather than just kicking

Interesting thanks for the tips so far. :)

Mineminer wrote: 3. How is "lag" measured in this game? I noticed that there on the "fresh" server a lag metric sitting at around 0.1~ when I joins but I have no idea how much (if any) lag that means? If that not lag then how "much" of this measure is "too much" to start being considered lag?

Lag is subjective, yet measurable.

It can be measured in many different ways. The engine exposes an `m_lag` value to the serverlist, and this has a specific meaning that is difficult to understand. In general, higher values mean longer delays between `cause` and `effect` which may affect game play, but nobody agrees on whether value X or Y is `terrible` or `just fine`.

It really depends on your player experience, in the end. Nothing will tell you better than players telling you that it's a problem, or just fine. Some players are on bad DSL connections, and they will always have lag everywhere. Some players are on dialup, and they will never have lag, so make sure to ask your players what kind of ISP/service they have if they complain, or praise your server.

What you can do, however, is collect and monitor data. If you see that m_lag varies wildly and has many spikes that are extremely high (e.g. above 0.5 or more, many times a day) then you may have a problem that you can solve. There are other values you can monitor, and people have written mods to monitor lag values in different ways to provide more data that can help you understand and take targeted action.

I see, thanks for the information on that regard.

I'm working on a mod that will give some more accurate and useful performance metrics than the "max_lag" used by the engine. For now, I wouldn't recommend relying on max_lag for any sort of benchmarking, since it doesn't represent maximum lag or even average lag for that matter. I have some maintenance routines in Lua that completely stall the server for upwards of 8 seconds, and the max_lag will still be reported as only ~0.3 seconds.

sorcerykid wrote:I'm working on a mod that will give some more accurate and useful performance metrics than the "max_lag" used by the engine. For now, I wouldn't recommend relying on max_lag for any sort of benchmarking, since it doesn't represent maximum lag or even average lag for that matter. I have some maintenance routines in Lua that completely stall the server for upwards of 8 seconds, and the max_lag will still be reported as only ~0.3 seconds.

Nice! Always gotta make those "lemonade" out of those "lemons" right? :)

sorcerykid wrote:I'm working on a mod that will give some more accurate and useful performance metrics than the "max_lag" used by the engine. For now, I wouldn't recommend relying on max_lag for any sort of benchmarking, since it doesn't represent maximum lag or even average lag for that matter. I have some maintenance routines in Lua that completely stall the server for upwards of 8 seconds, and the max_lag will still be reported as only ~0.3 seconds.

Nice. Are you going to make a git tree and do small incremental releases? Or just dump a bunch of zip files without history and post an incomprehensible changelog with confusing features? I'm hoping this time it's the former! :D