Planning a Laravel upgrade starts with the work your business must keep doing. List those tasks before you approve code changes. Then ask for a tested release plan, with a named person who can stop it.
A newer framework can help keep an application within support. It will not fix every fault in the business process. Keep those two aims separate when you agree the scope.
Do you need an upgrade or a wider rebuild?
An upgrade moves the existing application onto a newer framework version. A rebuild changes more of its design. Neither is the right answer just because the software is old.
If staff can do their work well, preserving that behaviour may be the goal. If they still need spreadsheets to finish each order, record that as a separate problem. Otherwise, a small upgrade can grow into a redesign without a clear decision.
Ask the team to split its findings into required changes and proposed improvements. Our legacy software modernisation page covers choices beyond replacing everything. Use a software specification to set out what must stay the same.
What must the team check before quoting?
Start with the versions in use, rather than the version someone recalls. Record Laravel, PHP, the database and hosting limits. PHP runs the application; Laravel is the framework built on it.
Laravel's release policy and PHP's support table have separate support windows. Check both for the current system and the proposed target. A hosting limit may rule out the planned change.
Packages are other code the application relies on. Some have their own fees, support rules or upgrade steps. Include payment, email and stock services too. Their test accounts and settings need an owner.
- Record the installed versions and proposed target.
- List packages, custom changes and paid licences.
- Check each outside service and its test setup.
- Identify code with no clear maintainer.
- Record known faults before work starts.
- Separate confirmed facts from items still to check.
The official Laravel upgrade guide describes changes for its target version. Ask which guides apply to your route. An automated upgrade tool can change code, but someone must still test the result.
How do you prove the business tasks still work?
Capture a baseline: a record of what happens before the upgrade. Save expected results for tasks such as a new order, a refund and a failed payment. Include user roles, totals and messages sent to other systems.
Our Easy Care Solutions project links a filter catalogue, subscriptions and courier cut-offs. Those linked functions show why an upgrade test must go beyond loading a page. Test the full route from a customer's choice to the dispatch task.
Developer tests should run alongside checks by staff who know the work. Our user acceptance testing guide explains how to set pass rules. Keep test results tied to the exact code version tested.
A package check is useful but narrower. Composer audit checks for known package security notices. It cannot prove that your own order rules are correct.
What should a safe rehearsal include?
Run the planned release on a separate test copy first. Use made-up or safely masked records. Block live email, payments and other outbound actions before background work can run. A copied system must not contact real customers by mistake.
Rehearse the steps in the order the live release will use. Record what changed and repeat the business tests. A rehearsal with missing jobs or fake success screens can hide the faults you need to find.
Include queue workers, which process work in the background. Include scheduled tasks, such as daily exports. Laravel's deployment guidance covers long-running processes that need to load the new code. The right steps depend on the installed version.
What does the fallback plan really recover?
Ask two questions before agreeing that the team can go back. Each needs a tested answer.
Can the old code use the changed data?
A migration changes the database structure or its contents. Old code may expect a field that the new release removed. Restoring old code alone may leave the application unable to run.
Have the team review each migration before release. The Laravel migration guide explains how schema changes are defined. Our planning advice is to test the whole recovery path. A reversal step alone does not prove that new records will be kept.
What happened after release?
Customers may have paid, received email or changed a booking. Restoring a database does not recall an email or cancel a payment elsewhere. It may also remove valid orders placed since the backup.
Agree how staff would find and check those events. Name who can approve recovery and who checks the affected records afterwards. Sometimes a controlled fix is safer than restoring an old copy. Decide from the actual fault and evidence.
Explore a fallback scenario
Choose what changed. Use the prompt to question the release plan, not to run a recovery.
Check whether old code can run
Reverting code may be an option. First check stored data, settings and outside effects. The release owner needs proof that the old version still works.
Check code and data together
Old code may not understand the changed data. Test the whole recovery path. A reversal command is not proof that records will be kept.
Protect valid new work
An old backup may lack new orders or bookings. Agree how the team will preserve and check those records before choosing a recovery path.
Review what happened elsewhere
A code change cannot unsend an email or undo a charge in another system. Name who checks those events and agrees the next action.
No commands run here. Your choice is not sent or saved.
What belongs in the release decision record?
Keep one short record with links to the test proof. The business owner should be able to read it without opening the code.
| Decision item | Record before release |
|---|---|
| Scope | Target versions, required changes and work left out. |
| Test proof | Code version, critical tasks, results and unresolved faults. |
| Data changes | Reviewed migrations and checks on records and totals. |
| Recovery | Tested path for old code, new data and outside events. |
| Stop rules | Failures that stop release and the person who can act. |
| Approval | Named business owner and any risk they accept. |
| Early live checks | Checks, evidence and owners for the first real tasks. |
Do not mark a blocked test as passed. Record the gap and its business impact. If the gap affects a critical task, resolve it before approving release.
Who checks the application after release?
Assign someone to watch the first real orders and background jobs. Check both sides of each vital link. A sent request does not prove that the other system saved it.
Check failed jobs, scheduled work, key totals and customer messages. Let staff know where to report faults. Keep the release owner available until the agreed checks are complete.
Then record who owns the next updates. A scoped upgrade ends; the need to maintain the application continues. Our website support and maintenance service is the next step when you need help with that ongoing work.