Out of Memory (OOM) Errors
Fix Out of Memory errors and prevent crashes on your Cafora server.
Illustrated procedure
Screenshots show the current application with demonstration data. Use your own service details; click an image to inspect it at full size.
Check the console
Open Essentials → Console. Check the current state, startup output and errors before making changes. Wait for Done or your software’s ready message before joining.

Locate the file log
Under Files, double-click logs and locate latest.log. Check modification times. Preserve the relevant file before a restart replaces or rotates it.

Read the complete error context
Open the log in the editor and inspect the earliest relevant error and surrounding lines. Remove tokens, passwords and private player details before sharing. Do not diagnose from the last line alone.

Describe the request to support
Open a support ticket with the service or invoice ID and the requested action. For connection issues include time, error and MTR evidence. Never include passwords, tokens or card details.

Preserve the error before classifying it
Save the time, timezone, exit information, game logs and recent changes. Avoid repeated restarts: termination can involve Java, container limits or another process failure.
| Symptom | Next check |
|---|---|
OutOfMemoryError: Java heap space | Check heap budget and workload; retain the full error, which does not necessarily mean a leak |
Metaspace or native allocation failure | Investigate the relevant pool and process budget instead of only increasing the heap |
| Sudden termination without a Java exception | Preserve Panel exit information and ask support to check container or host events; an exit code alone is not conclusive |
Recovery sequence
- Preserve logs. Stop the server before file changes and back up configuration and saves.
- Review recent server, plugin, mod and startup changes. Revert a known change first.
- Check the JVM memory budget. Increasing
-Xmxinside a constrained container can reduce native-memory headroom. - When appropriate, attempt one controlled start and observe the same workload with spark.
Do not delete worlds, dimensions or mods merely to free RAM. Disk size does not measure loaded data, and removing mods can affect save compatibility.
Support handoff
Include server identifier, time and timezone, complete error context, recent changes and resource charts. Remove secrets first. Discuss optimization or capacity changes after establishing the actual pressure. Scheduled restarts are not a substitute for diagnosis.
Reference: Oracle’s memory-error diagnosis. See the log guide for the console and file walkthrough.