Power input, PSU, fuse
No AC input, tripped breaker, PSU in fault with an amber LED, power cables not seated, a smart socket that fails to re-arm after an electrical event. Most frequent cause: check the power supply before touching the hardware.
By manufacturer
Dell PowerEdge HPE ProLiant Lenovo ThinkSystem Fujitsu Primergy Supermicro IBM System x / Power Acer Altos
Repair · no POST
Server won't boot: progressive isolation. The cause is often trivial.
No POST, blank display, an error code on the LCD panel, a solid amber LED: the "won't power on" symptom typically has a cause you can pin down in a few steps. Beep code decoding, readout of the vendor diagnostic LEDs, progressive isolation by category (power → CPU → RAM → storage).
The process
Five categories of cause, in order of frequency.
On enterprise servers the "won't boot" symptom covers very different causes. We work down in order of decreasing probability, starting with the most trivial causes — the ones nobody thinks of:
Diagnostic codes, beeps, LEDs
Vendor diagnostic LEDs pointing to CPU fault, motherboard fault, VRM fault. Codes on the LCD panel (PowerEdge: CPU0001, MEM0801; ProLiant: AMP indicators). POST beep codes. Often identifies the culprit unambiguously.
DIMMs not detected or in fault
POST hanging at memory detection, memory beep codes, DIMMs disabled by the BIOS after earlier uncorrected errors. Diagnostic test: pull every DIMM except one (in the vendor-recommended slot) and boot incrementally.
POST passes but OS boot fails
POST completes but boot fails: RAID controller not importing the array, no boot device selected, corrupted MBR/GPT. Distinct from a pure "no POST" because the system at least reaches the vendor splash screen.
iDRAC/iLO/XClarity hung
A special case: the system appears "unresponsive" but the BMC is actually in an abnormal state. Reset the BMC (vendor-specific key combination, or a full power cycle with AC removed as well), then retry.
External device blocking POST
A USB port with a faulty device attached, an unusual KVM cable, a dongle presenting itself as an unrecognised boot device can all stall POST. Disconnect everything and attempt a bare boot: a cheap step that resolves the problem surprisingly often.
FAQ
The questions we get most often.
The server was running yesterday, today it won't start. What do I check first?
1) Power cable physically seated and UPS / smart socket working. 2) Front LEDs and LCD panel: is there an error code? 3) If no LED is lit at all, it is probably a power problem (socket, breaker, PSU). 4) If the LEDs come on but the system doesn't boot, the LCD panel or the AMP/diagnostic LEDs point to the category. Document every LED before touching anything.
Can I read the error code without a monitor attached?
Yes, in two ways: the front LCD panel on models that have one (PowerEdge R-series, ProLiant DL-series), the BMC web interface via iDRAC/iLO/XClarity/IPMI/iRMC/CIMC if reachable, under System Event Log or Boot Progress. The BMC works even when the operating system isn't loaded — the server only needs standby power.
Do you swap parts by trial and error if you can't find the cause?
No, never. If the diagnosis doesn't isolate the culprit with enough confidence, we say so and propose intermediate steps (e.g. a minimal configuration test in the lab). Swapping components by trial and error is the fastest way to spend a lot without fixing anything and — worse — to introduce new problems.
How long has the symptom been there? Does it matter?
A great deal. A server that has never booted since manual work was done on it (e.g. opened for cleaning) suggests a problem introduced by that work (cable not reconnected, DIMM not fully seated, screw dropped on the motherboard). A server that shut down suddenly under load suggests a power or thermal issue. A server that won't boot after a long shutdown suggests capacitors or the CMOS battery. The context drives the diagnosis.