Maintain a FiveM script by keeping a reproducible version, testing its real gameplay flows and retaining a working rollback. A resource starting without errors is only the first check; permissions, persistence and failure paths matter just as much.
Record the working baseline
List the resource version or commit, FXServer build, framework, inventory, database driver and configuration changes. Keep editable source in a private repository and credentials outside it. Preserve the original licensed package and the author’s update instructions.
Use an isolated test server with a separate database and restricted access. Disable real payment delivery, production webhooks and other external actions in the test environment.
Review each update before installing
- Read the release notes and compare changed configuration, dependencies, exports, events and SQL. Identify breaking changes before replacing files.
- Back up the current database and matching resource/configuration files. Confirm that the backup can be restored on the test instance.
- Apply the update to the test instance in the author’s required order. Merge configuration deliberately; do not blindly overwrite the new defaults with an old file.
- Run normal and failure scenarios, then compare persisted results after reconnect and restart.
Use a gameplay test record
For each important action, record the starting state, steps, expected result and observed result. For a shop purchase: the player has a known balance, buys one known item, receives exactly one item, loses exactly its server-defined price and keeps that result after reconnect.
Repeat with insufficient funds, full inventory, unauthorized access, invalid item names, repeated requests and a disconnect during the operation. For a garage, test ownership, duplicate spawn prevention and storage after restart. For jobs, test rank and duty restrictions.
Inspect security and performance separately
Review network events for server-side validation of permissions, ownership, amounts and task completion. A critical authorization failure blocks release regardless of good performance elsewhere.
Use the client resource monitor for client script timing and the FXServer profiler for server work. Record the scene, player count and duration. Compare identical workloads; do not impose a universal millisecond budget on unrelated resources.
Automate checks that catch real mistakes
Run Lua syntax/lint checks suitable for the FiveM runtime and focused tests for pure logic. Mocked framework tests help detect argument and return mismatches, but they do not verify the actual framework, inventory or database integration. Keep a live test-server checklist alongside them.
Deploy a reversible change
Choose a maintenance window, announce the expected interruption and keep the previous release ready. Update only the reviewed package and required dependencies. Restart in the documented order, check logs and repeat the key gameplay scenario.
If verification fails, stop new writes before recovery. Restore the matching code and any necessary data, accounting for changes made after the backup. Do not silently wipe player progress to make a test pass.
Keep an issue useful
Record the first relevant error, affected resource/version, reproduction steps and expected behavior. Redact credentials and personal identifiers from logs. Track the resolution in the resource’s change history so the next update can repeat the same regression check.