Use

Crashes and health

See CPU and memory over time, every crash with a timeline, and a written report on what most likely happened.

Turning it on

Crash reports and the health monitor need Server health in your data choices. On new servers it is on by default. On older servers the settings page shows a banner: Turn on Server health to see crashes, CPU and RAM. One click is enough. Only the owner can turn it on.

You find everything under Crashes (the pulse icon in the settings header) and under Health in the server settings.

The health monitor

Health shows CPU, memory and disk of the machine your server runs on, and how much of it FXServer uses. Switch between the last 24 hours, 7 days and 30 days. Hover or tap the chart for exact values.

Look for memory that only goes up, or CPU that stays near the top. Both often come before a crash.

The crash list

Crashes lists every crash, freeze and restart loop, newest first, each with a timeline and a report. A new entry shows Collecting first, then Analyzing while the report is written.

Kind	What it means
Crash	FXServer crashed.
Crash, process kept running	FXServer crashed, but its process did not end. Common in containers.
Server froze	The server stopped responding, often because a script got stuck.
Out of memory	The process ended while memory was nearly full.
Server stopped	The process ended without a restart from txAdmin or an admin.
Restart loop	Several crashes or restarts in a short time.
Manual check	You pressed “My server just crashed”.

Crash
FXServer crashed.
Crash, process kept running
FXServer crashed, but its process did not end. Common in containers.
Server froze
The server stopped responding, often because a script got stuck.
Out of memory
The process ended while memory was nearly full.
Server stopped
The process ended without a restart from txAdmin or an admin.
Restart loop
Several crashes or restarts in a short time.
Manual check
You pressed “My server just crashed”.

Crash reports are kept for 180 days.

One crash, step by step

A crash page reads from top to bottom:

Headline and facts: what happened and when, with players online, how long the server had been up, FXServer memory and money rolled back.
Timeline: CPU, memory and players on one time axis, then server events, log lines by type and money moves. Switch between 1, 4 and 12 hours around the crash.
Rollback window: money that moved after the last save before the crash. It was lost in the restart, or may now exist twice.
Crash dump: what FXServer's dump file says, in plain words.
Report: what happened, likely causes with the evidence for each, the impact and what to do next.

CrashStarted againScript errorsHitchesRollback window

How a crash page lays out what happened: every panel shares one time axis, and the red line marks the moment of the crash. Illustration only.

Without txAdmin

A server without txAdmin writes no log file. The timeline then shows the process, CPU, memory and crash dumps, but no log lines.

“My server just crashed”

Players say the server is down, but nothing is in the list yet? Press My server just crashed on the Crashes page.

If a crash was found in the last 30 minutes, it opens that one.
Otherwise the agent checks the last 30 minutes right away: the server log, the process and any new crash dumps. The page updates by itself.

You always get a page back, even if it says no crash was found. It still shows what was seen, and you can ask about it.

The crash assistant

Under each crash you can ask follow-up questions, for example “Which resource caused this?”, “Has this happened before?” or “Did anyone get money out of the rollback?”.

The assistant reads the timeline, the server log, the crash dump and the money moves of that crash. Answers cite times and numbers from your server's data. When the data does not say, the assistant says so.

Crash dump cleanup

FXServer writes a dump file for every crash into its crashes folder. Big servers collect gigabytes of them over time.

Under Health, turn on Delete old crash dumps and choose how many to keep (1 to 100). The agent keeps the newest ones and deletes older ones, together with their large full dump. It only deletes a dump after reading it. Off by default. Owners and admins can change this.

The large dump stays with you

Only a cleaned copy of the small dump is sent, with passwords and keys removed. If a secret cannot be removed, the file stays on your server and only the finding is sent. The large full dump never leaves your machine.