The plugin list is where new servers go wrong, and the mechanism is not the one people expect. It is not that any single plugin is bad. It is that thirty plugins from twenty authors is thirty independent chances that the next game update breaks your server, and twenty independent chances that the person who maintains one of them has stopped.
What you actually need on day one
Four things, and none of them is exciting:
- Permissions, because every other plugin will depend on it.
- A ban system with persistent records, because the built-in one forgets when the file is edited by hand.
- World protection at spawn, unless your gamemode is explicitly destructive.
- A backup plugin or script, verified by restoring once.
That is a working server. Everything past this point is a choice rather than a requirement.
The permissions plugin is a bigger decision than it looks
Every other plugin integrates with it, so changing it later means reconfiguring all of them. Pick the one your intended plugin set actually supports and stay with it.
The failure people hit is choosing based on a tutorial, discovering it does not integrate with something they add in month three, and running two permission systems side by side. Two systems that both think they own permissions produce the state where somebody has access in one and not the other, and nobody can work out which is authoritative.
That is the same shape as any other split-brain configuration, and it is why admin systems deserve deliberate design rather than accretion.
What to resist
Anything that duplicates something you already have. Two chat plugins, two economy plugins, two anti-grief plugins. They will conflict, and the conflict will be subtle.
Plugins with no update in two years. They may work perfectly today. They will not survive the next major version, and you will discover that on update day with players waiting.
Anything that advertises "all-in-one". A plugin that does chat, economy, homes, warps, kits and moderation is a plugin whose bugs affect all six, and whose abandonment costs you all six at once.
Cosmetic plugins, early. Particles, titles, join messages. They are the most fun to install and the first thing that breaks, and none of them is why anybody stays.
The update-day arithmetic
When the game updates, every plugin has to catch up. A server with five plugins can usually update within a day. A server with thirty waits for the slowest of thirty authors, and there is always one.
That wait is not a technical problem — it is a community one. A server that is two versions behind for a month because of one unmaintained plugin bleeds players to servers that updated, and the plugin that caused it is usually something nobody would have missed.
Count your dependencies before you add one. The question is not "is this plugin good" but "am I willing to wait for this author, every time, forever" — the same calculation that makes a Source server's plugin list a liability rather than a feature.
Where the plugin came from matters as much as what it does
A plugin downloaded from a public repository with a visible commit history and an issue tracker is a different proposition from a jar posted in a Discord channel, even when the code is identical.
The difference is what happens in six months. The first one has somewhere to check whether it is still maintained, somewhere to read what changed, and somewhere to see whether anyone else hit the bug you just hit. The second has a person who may or may not still be in that Discord.
Server owners install the second kind constantly, because it is usually free and usually solves the exact problem in front of them. It is a reasonable trade for something cosmetic and a poor one for anything holding player data, and anything holding player data is also the thing you cannot simply remove later.
Add them one at a time
Install. Restart. Play for a day. Then the next.
A batch of eight new plugins produces errors that interact, and separating them afterwards costs more than adding them slowly would have. This is the single most-ignored piece of advice in server administration and the reason so many consoles are full of warnings nobody can attribute, which is exactly what makes a real crash impossible to trace when one finally arrives.
The list that ages well
Small, maintained, and boring. Servers that are still running in three years tend to have fewer plugins than they did in year one, because the owner removed things rather than adding them.
The removal is the hard part. A plugin that six people use is a plugin six people will complain about losing, and keeping it is how a server ends up unable to update, which is a slower and more permanent version of the decline nobody notices until the population has already gone.
- Tags
- pluginsadministrationmaintenance
- Share
Published · 4 min read