Memory hook: Reproduce the symptom without destroying the evidence.
Must remember
Separate a single application, one user profile, the OS and shared services. Compare a working user/device and inspect recent updates, permissions, storage, dependencies and network paths. Record exact errors and timestamps; “it is slow” needs measurable context.
Crashes can involve application defects, incompatible extensions, corrupted profiles, missing runtimes, insufficient resources or hardware faults. Use Event Viewer/reliability data on Windows, platform logs elsewhere and safe vendor diagnostics. A clean boot/Safe Mode can isolate startup components, but restore the intended configuration afterward.
Boot loops and failed updates may require recovery tools, driver rollback, repair or restoration. Preserve data and recovery keys before destructive actions. A BSOD/stop error is evidence to investigate, not an instruction to reinstall immediately. Low disk space can block updates, paging and application writes.
Mobile issues include app permission denial, full storage, synchronization conflicts, network switching, battery optimization and management restrictions. Try scoped steps such as checking settings, updating/restarting the app and verifying the service before wiping the device.
Browser failures can originate in DNS, proxy, certificates/time, extensions, cached state or the remote service. Check whether other sites/users work and whether the issue occurs in a controlled clean profile. Do not teach users to bypass certificate warnings as a routine fix.
After repair, retest the same workflow, verify security controls and document changes. Escalate with useful evidence when the cause belongs to a backend service or application owner.
Choose under exam pressure
| Requirement | Choice and reason |
|---|---|
| Only one profile affected | Investigate profile-specific settings and permissions. |
| All users fail simultaneously | Check shared service/network/change dependencies. |
| Certificate warnings suddenly appear | Check time, endpoint identity and trust chain before proceeding. |
Traps
- A reset can hide the cause while losing data.
- Disabling security permanently is not an acceptable generic performance fix.
Active recall
1. Why compare a working user or device?
It isolates shared versus local causes.
2. What makes an error report useful?
Exact message, time, scope, reproduction steps and recent changes.
3. Why can low free disk space cause varied failures?
Updates, temporary files, paging and application writes all need space.
4. Should certificate warnings simply be ignored?
No. Verify time, identity and trust rather than bypassing the protection.
5. What proves the fix?
The original workflow succeeds with required security and no unacceptable side effects.