Bukkit

Let Me Explain

PluginMetrics is an important tool, both to server owners and plugin developers. If I were limit my Minecraft-related donations to exactly one plugin that donation would go to PluginMetrics. Given the kind of questions I've seen about this on the Bukkit forums it is clear to me that many server administrators don't know what PluginMetrics does or why it is such a valuable tool to server administrators and plugin developers.

In brief, PluginMetrics collects anonymous details about a game server and submits them to a central database (http://mcstats.org/). I want emphasize the anonymous part. This means that the database does not store or display information about specific servers. So it is not possible for me to go to mcstats.org, search for MwanzoCraft (the server I play on the most), and find out anything about MwanzoCraft. That's not the goal of PluginMetrics, there are MANY "Minecraft server listing sites" that list that kind of information if the administrators of a game desire to be published in that fashion.

The goal is collect the kind of information that helps plugin developers (authors, if you will) make intelligent decisions about their plugins. Here's some of the questions that can plague a plugin author, and PluginMetrics can help answer:

• Which Java platforms should we support?
• Which versions of Minecraft should we support?
• Do we still need to be compatible with Minecraft 1.2.5?
• ...and many many more.

It allows plugin authors to decide intelligently, with real data, whether to continue supporting a given version of the game server, whether to continue supporting the plugin on Java 1.6, etc. Without those metrics to tell us a given platform or game version, some servers could get left out in the cold, because their version of the game is dropped from the plugin's support.

For instance, as of this writing, the game version break down for CommunityBridge is:

• 1.6.2: 55.60%
• 1.6.1: 1.90%
• 1.5.2: 29.63%
• 1.5.1: 3.7%
• 1.4.7: 7.4%
• 1.4.6: 1.9%

I can learn a great deal from these numbers. For instance, I can't yet remove the code that makes CommunityBridge back-compatible with 1.4.x, because almost 10% of CommunityBridge servers are still on that version of Minecraft. I also tells me that even though there is no beta or recommended build for 1.6.2 at this time, more than half of CommunityBridge servers are already on that platform (and apparently without any problems, since I've not had any error reports). That makes rigorously testing CommunityBridge against 1.6.2 a high priority.

Likewise, I was very surprised recently to learn this breakdown:

• CraftBukkit: 44.44%
• Spigot: 40.74%
• MCPC+: 14.81%

Which tells me, if I can make the time, there's benefit to testing against Spigot and MCPC+.

Let Me Sum Up

The take away here is, if you disable metrics, your server isn't counted. And you want your server counted, because you want your server's Java Platform, your server's specific version of CraftBukkit (or Spigot, or MCPC+), your chosen Permissions System, to continue to be supported. If no one using your particular style/version/etc. of server is counted, then plugin authors like me will choose to devote our extremely precious time to other versions, platforms, permissions systems, etc. And that means YOUR server's preferred setup won't be supported. So right now, go to the plugins/PluginMetrics/config.yml file and set opt-out to false. Using pluginmetrics will make your favorite plugins better and that will make your server better.

Take the time to actually look at the data there. Take a look at the global stats and compare it to the stats of your favorite plugin. How do the numbers compare? Is your favorite also the "majority's" favorite? How many of those servers have already updated to 1.6.2? etc. Think about the ramifications of those numbers...

Specific Myths

"Only" How Many AND "A download count can do just as well"

1. Download counts are NOT a good indicator of how many servers are using a plugin. I've downloaded dozens of plugins I never ended up using. I've also downloaded the exact same release of a given plugin several times. Download counts, in other words, have MANY false positives included in their count.
2. A download count isn't real time data. So any server that downloaded the plugin, used it for "some amount of time" and then abandoned it is counted by a download count, but not by PluginMetrics.
3. Finally, PluginMetrics gathers many other points of data, most of which are more useful than knowing "how many servers are using the plugin." It is much more useful to know things like "80% of my plugin's servers are running CraftBukkit 1.6.2."

Lag

Sometimes, when searching for causes of lag on a given server, an administrator will conclude that PluginMetrics is the culprit. This is very unlikely. The update task runs asynchronously and is very light. I strongly recommend you look closer at other plugins (in particular, I'd look for plugins that have block place/break listeners), other features of the game, your hardware, or your network connection for causes of lag.

References

1. Plugin Metrics Bukkit Thread
2. MCStats (the database itself)