Check, isolate, verify

Troubleshoot with Evidence

Fix Mini PC and home server issues with measured, reproducible evidence: Wake-on-LAN, Docker, Jellyfin, NVMe and Windows diagnostics that actually verify.

Every fix article here follows the same workflow: reproduce the problem, collect measurements before touching anything, change one variable at a time, and verify the fix with the original workload rather than a quick smoke test. We keep failed runs and dead ends in the write-ups because the boundary between "works" and "does not work" is usually more useful than the fix itself.

If you are rebuilding a Mini PC after a BIOS or driver change, start with the Wake-on-LAN sleep vs shutdown fix — it exposes the difference between S3 and S5 states that most Mini PC BIOSes hide. For storage builders running Docker on Windows, the Docker reboot recovery article walks the full dependency chain from Windows login through WSL2 to the Compose stack. Media server operators hitting "QSV should work but Jellyfin says software" should read the hardware-acceleration fix before buying a GPU.

More fixes are added as we reproduce and verify them. If you hit a reproducible issue on supported hardware, open a discussion and we will run the workflow.

Memory compatibility

Mini PC Memory Compatibility: DDR4 SO-DIMM Capacity, Rank and Voltage

Generation, voltage, rank and per-slot capacity decide whether a module will post. Our measured EQi12 configuration plus an owner-report checklist.

Read article
Power limits

Mini PC PL1 and PL2: Why a Slow Mini PC Is Not Always Hot

How firmware power limits cap sustained performance with no heat at all, how to tell a power wall from a thermal wall, and what our 37 W load reading does not prove.

Read article
Thermal

Mini PC Thermal Throttling on a Home Server: How to Prove It

The three protection layers, the evidence pattern that proves throttling, and why our own 12.57-hour run peaked at only 52% CPU.

Read article
Fan policy

Mini PC Fan Always Loud: BIOS Fan Policy Explained

Why a fan ramps up with no load, what the Smart Fan stop, start and full-speed thresholds actually do, and the order to tune them in.

Read article
Wake and dropout

USB and Network Devices Missing After Sleep on a Mini PC

Selective suspend, root-hub power flags, Fast Startup and hub current budgets, with our measured 4/5 sleep-wake result.

Read article
Windows power states

Windows Fast Startup and Wake-on-LAN: Was Your Shutdown Really S5?

Fast Startup turns shutdown into an S4 hybrid. Verify the real power state with powercfg, event logs and our measured five-cycle S5 and S3 results.

Read article
Wake-on-LAN

WOL Works from Shutdown but Fails from Sleep

Separate S5 from S3, map the active Realtek adapter and turn an intermittent 4/5 result into a repeatable diagnostic.

Read article
USB storage

USB SSD Is Limited to About 40 MB/s

Isolate a USB 2.0-like bottleneck with port mapping, negotiation checks, repeated transfers and readback hashes.

Read article
Docker

Docker Containers Do Not Start after Reboot

Check Windows, login, WSL2, Docker, Compose, container health, application readiness and persistent data in order.

Read article
Jellyfin

Jellyfin Still Uses Software Transcoding

Distinguish Direct Play, hardware decode, filters and encode; prove Intel QSV from the complete FFmpeg pipeline.

Read article
Windows image

Two OEM Mini PCs Share a Machine SID

Handle duplicated factory-image identity with private evidence, a clean installation and carefully scoped publication language.

Read article
Wake-on-LAN

Wake-on-LAN Not Working: 12 Causes Ranked

Probability-ordered checklist across firmware, driver, power state, link speed and network path, with a 5-run test protocol.

Read article
ACPI states

Wake from S5: PME, RTC Alarm and Wake-on-LAN

Which wake source works in S3, S4, S5 and G3, with measured 5/5 S5 Wake-on-LAN and 42 s service recovery.

Read article
NVMe health

NVMe Critical Warning 0x02 Temperature

Decode the smartd temperature alert, read the two cumulative temperature counters and spot a stuck daemon state or a bad sensor.

Read article
RAM testing

Windows Memory Diagnostic on a Home Server

Schedule the test unattended with bcdedit, then read the real pass/fail verdict from event 1101, with a measured 16 GB run as the reference.

Read article