Lenovo Support in Taiwan After End of Life

Several Minutes of Nothing Is Sometimes Correct

A Lenovo server with a large memory configuration performs memory initialisation and testing before it produces any video output. On a heavily populated machine this can take several minutes during which the display stays blank and the machine appears unresponsive. Taiwanese operators reasonably read that as a failed startup, and the usual response — cutting power and trying again — is the one action that guarantees it never completes.

Each Interruption Restarts the Test

The memory pass begins again from the start after a power cycle. Three impatient restarts produce three incomplete passes and a machine that has never reached an operating system, which is then reported as a hardware failure. We are asked to quote system boards for machines in exactly this state with some regularity, and the first question we ask is how long the machine was actually left alone.

The management controller is the instrument

The display is the wrong place to look during startup. The management controller reports which phase the machine is in and how far through it is, and it is reachable over the network while video output is still blank. Consulting it converts an anxious guess into a known state, and it is the single habit that prevents this class of unnecessary call-out.

A slot is not a module

Where there is a genuine memory fault, the machine either maps out the failing module and continues with reduced capacity or halts with a reference to a specific slot. That reference is a location, not a diagnosis. A failing memory channel on the system board presents as a fault in whichever module occupies it, so swapping the module resolves nothing and the same slot reports again. The test that separates the two is moving a known-good module into the reported slot and seeing whether the fault follows the module or stays with the location.

What ICD Supplies for Lenovo Estates in Taiwan

ECC modules matched by rank and generation to the bank they are joining rather than merely by capacity, system boards for cases where the channel rather than the module is at fault, RAID adapters with the cache hardware their generation requires, and hot-swap power supplies for chassis the current channel no longer lists. We quote against machine type and model. All hardware is tested against a like machine, serial-logged, and carries one to three years of cover.