Guide · Updated 14 Sep 2026
RedM server performance: where the lag comes from and how to get it back
"The server is lagging" describes four different problems, and the fix for one makes another worse. Before changing anything, work out which one you have: a client burning frames on a script, a world filling up with entities, a database doing too much, or a server process that is actually out of headroom. This guide takes them in the order they usually turn out to be the cause, with the settings to change on the scripts that matter.
Step 0: measure
Two tools, both already installed.
- resmon, typed in the F8 console on a client, lists every resource with its CPU time per frame and memory. Sort by CPU. This is where client lag shows up, and it points at one script by name.
- txAdmin's dashboard shows server thread usage, player count and tick time. Server-side problems show here as a rising tick time; if it is flat and players still report stutter, the problem is on the client.
Write the numbers down with the player count. A change you cannot measure is a change you cannot keep.
Cause 1: a client script in a tight loop
The most common cause, and the one with few players. A script that checks something every frame, distance to every prompt, every player's position, whether a key is held, costs a slice of every frame on every client. Ten of them and the game stutters at 20 players. In resmon they sit above 0.10 ms while nothing is happening.
What to do: the script's config often has an interval. Raise it. Admin Blips (Free) updates every player's blip every 500 ms by default; that is fine for admins but there is no reason to run it at 100. Scripts with no interval and a high idle cost are worth an author's attention, or a replacement.
Cause 2: entity build-up
Dead horses in the road, wagons abandoned at a station, props a script spawned and never deleted. Each one is simulated by every client near it, and RDR2 does not clean up as fast as a busy server makes mess. This is the cause of "it was fine at restart and awful three hours later".
What to do: schedule cleanup. Poggy Util (Free) has object removal on a timer for the models you list, and a dead entity cleanup that clears dead NPCs, horses, animals and wagons, with an admin command to run it now. The cleanup starts switched off; turn it on and set the interval to fifteen or twenty minutes. Restarts every six to eight hours are the other half of the answer, and txAdmin schedules them.
Cause 3: NPC systems tuned too high
Any script that spawns NPCs with behaviour, witnesses, ambient law, bandits, adds entities that clients simulate and events the server relays. They are usually fine at their defaults and awful when an owner turns every number up. Witnesses (from $6.99/mo) caps the number of active witnesses and the search radius for this reason, has cooldowns per crime, and lets you blacklist towns like Sisika and Van Horn so prison roleplay and warzones stay quiet. Keep the caps, and lower the trigger chances before you raise the radius.
Cause 4: the database
Reads are rarely the problem; writes and bad queries are. A script that saves a character every few seconds, or one that runs a query over a table that has grown for a year on a column with no index. oxmysql logs slow queries; a query that takes more than a few hundred milliseconds during play is worth finding the owner of. Scripts that create their own tables on start, as every Poggy script does through Poggy Core (Free), ship the indexes with them; hand-imported SQL from an older release may not have.
Two habits: keep the database on the same machine as the server, or on the same network with a low ping, and back it up nightly so a bloated log table can be trimmed without fear. See how to back up a RedM server.
Cause 5: the server itself
Last because it is rarest. If tick time on the dashboard climbs with player count and stays high, the machine is short of a fast core (RedM's server thread is single-core bound) or short of memory. The fix is hardware, but check the four causes above first: a server that is doing three hours of dead-entity bookkeeping looks like it needs a better CPU.
Settings worth changing today
| Where | Setting | Suggestion |
|---|---|---|
| txAdmin | Scheduled restarts | Every 6 to 8 hours, announced in advance |
| Poggy Util | Dead entity cleanup | On, every 15 to 20 minutes |
| Poggy Util | Object removal | On, with the prop models that block your roads |
| Witnesses | Max active witnesses, search radius | Leave at default; lower trigger chances instead of raising radius |
| Admin Blips | Update interval | 500 ms or slower |
| Any script | Debug logging | Off. Console spam costs more than it looks |
What does not help
Raising the server's tick rate, adding "optimization" resources that promise to fix everything, or moving to a bigger host without measuring. Each has made a server worse that we have been asked to look at. Measure, find the one script or the pile of entities, and fix that.
