Fix Lag on Your Minecraft Server: How to Get 20 TPS

Two numbers you need to know: TPS and MSPT

• TPS (ticks per second) is how many times per second the server updates the world. 20 is the maximum - the server can never run faster, only fall behind.
• MSPT (milliseconds per tick) is how long one tick took. The budget is 50 ms. When a tick takes longer than that, TPS drops.

MSPT is the better metric. A server at 45 ms and one at 5 ms both report 20.00 TPS - but the first one is one mob farm away from collapsing. TPS tells you it already went wrong; MSPT tells you how close you are.

---

Step 1 - Measure before you change anything

Never guess. Run the measurements first:

Command	What you get
/tps	TPS averaged over 1, 5 and 15 minutes
/mspt	Average, minimum and maximum MSPT
/paper mobcaps	How close the world is to the mob cap
/paper chunkinfo	Number of loaded chunks

spark is the real tool, and it is bundled with Paper from 1.21:

/spark tps
/spark health
/spark profiler start --timeout 600

The profiler runs for 10 minutes and gives you a link to a report showing exactly which plugin or task is spending the time. If you only get occasional spikes, add --only-ticks-over 50 and capture just the slow ticks.

Timings is deprecated and disabled by default in newer Paper versions. Use spark.

---

Step 2 - Lower simulation-distance (the single biggest win)

In server.properties there are two settings people constantly mix up:

Setting	Default	Recommended	What it controls
view-distance	10	7	How far players can see. Costs bandwidth
simulation-distance	10	4	How far the server thinks: mobs, redstone, crops. Costs CPU

simulation-distance is what costs you TPS. Dropping it to 4 is barely noticeable to players - they still see far, the distant world simply stands still. Simulation distance is meant to be lower than view distance.

---

Step 3 - Get mobs and entities under control

bukkit.yml in the server root:

Key	Default	Recommended
spawn-limits.monsters	70	20
spawn-limits.animals	10	5
spawn-limits.ambient	15	1
ticks-per.monster-spawns	1	10
ticks-per.water-spawns	1	400

Two things get misunderstood here: the limits are multiplied by the player count, so 70 monsters with 40 players is an enormous number. And ticks-per is the interval between spawn attempts - a higher number means less work.

spigot.yml in the root folder:

Key	Default	Recommended
entity-activation-range.animals	32	16
entity-activation-range.monsters	32	24
entity-activation-range.misc	16	8
entity-activation-range.flying-monsters	32	48
mob-spawn-range	8	3
nerf-spawner-mobs	false	true
tick-inactive-villagers	true	false

entity-activation-range decides how far from a player a mob still gets AI and pathfinding. It is the most effective lever after the distance settings. Do raise flying-monsters, though, or phantoms and ghasts freeze in mid-air.

Keep mob-spawn-range at or below your simulation-distance.

---

Step 4 - Items, hoppers and redstone

In spigot.yml:

merge-radius:
  item: 3.5
  exp: 4.0
ticks-per:
  hopper-check: 8

In config/paper-world-defaults.yml:

Key	Recommended	Effect
entities.spawning.alt-item-despawn-rate.enabled	true	Junk like cobblestone despawns faster
entities.armor-stands.tick	false	Armor stands stop costing tick time
entities.behavior.update-pathfinding-on-block-update	false	Large win on servers with many mobs
hopper.disable-move-event	true	Big win - only if no plugin listens to InventoryMoveItemEvent
collisions.max-entity-collisions	2	Less work when mobs stack up
misc.redstone-implementation	ALTERNATE_CURRENT	Noticeably faster redstone
chunks.prevent-moving-into-unloaded-chunks	true	Stops elytra players outrunning chunk generation

---

Step 5 - RAM and Java

• Set -Xms and -Xmx to the same value. A growing heap creates GC work, and GC work is a lag spike.
• Do not allocate all your RAM. Java needs memory outside the heap too. With 8 GB available, give the server around 6.5-7 GB.
• Use Aikar's flags - the recommended JVM setup from PaperMC. The current string is at docs.papermc.io/paper/aikars-flags.

Java version by server version:

Minecraft version	Java
1.17 - 1.19	Java 17
1.20 - 1.21.11	Java 21
26.1 and newer	Java 25

More RAM does not fix low TPS. TPS is a CPU problem. RAM only helps if /spark health or /spark gc shows real memory pressure.

---

Step 6 - Pregenerate the world

Chunk generation is the most expensive thing a Minecraft server does. If your world is not generated in advance, you pay that bill every time a player flies somewhere new.

Install the Chunky plugin and run:

/chunky world world
/chunky center 0 0
/chunky radius 5000
/chunky start

Run it while the server is empty - the generation itself eats all the performance while it works. Afterwards, set a world border so players do not simply fly outside it again.

---

What does not help

• More RAM on its own. Only buy more if the measurements show memory pressure.
• /reload. Java cannot safely reload code. The command leaves old listeners and scheduled tasks running alongside the new ones. Restart instead.
• More CPU cores. The server tick loop is largely single-threaded - clock speed matters more than core count.
• Pasting a ready-made "optimised config". Several popular tweaks change gameplay. Fix what the measurement points at.

---

Troubleshooting

Symptom	Likely cause	Next step
TPS drops when many players are online	Simulation distance too high	Set it to 4 and measure again
TPS drops at fixed intervals	Autosave or a plugin task	/spark profiler start and see what runs
Lag spikes but a healthy average	Individual long ticks	/spark profiler start --only-ticks-over 50
The server lags when someone flies	Chunk generation	Pregenerate with Chunky
TPS decays over time after a restart	Entity or item build-up	/paper mobcaps, check your farms
Constantly low TPS regardless of players	A plugin is blocking the main thread	spark report, remove plugins one at a time

Cannot read your spark report? Send us the link in a ticket and we will go through it with you.