By manufacturer
Dell PowerEdge HPE ProLiant Lenovo ThinkSystem Fujitsu Primergy Supermicro IBM System x / Power Acer AltosBackplanes and connectors: errors that look like drive failures but aren't.
Drives dropping intermittently out of the RAID, SMART errors that migrate from one slot to another, NVMe devices that aren't always enumerated at boot: often it isn't the drives, it's the backplane or a connector. Targeted diagnostics and repair or replacement of the component actually at fault, without swapping out healthy drives.
A "strange drive error" whose behaviour keeps shifting.
The classic pattern that points to the backplane or a connector rather than the drive:
- Drive dropped from the RAID with Predictive Failure events, then coming back on its own after a reboot.
- The error follows the slot: replacing the drive with a new one in the same slot brings the problem straight back.
- CRC errors in the RAID controller logs — a sign of trouble in the electrical path, not on the drive media.
- NVMe not enumerated intermittently at boot: degraded U.2 / M.2 connector pins, a loose MCIO / SlimSAS cable.
Diagnostics, replacement, tidy recabling.
- Physical inspection: backplane pins, connector condition, any signs of corrosion or oxidation.
- Diagnostic swap: the drive goes into another slot, the drive from that slot goes elsewhere. If the error follows the slot, it's the backplane.
- Backplane replacement with an OEM or equivalent part, accounting for firmware variants where relevant.
- Internal recabling: SAS/SATA/NVMe cables replaced, routing optimised to reduce mechanical stress, labelled for future servicing.
The questions we get asked most.
How do you tell a drive failure from a backplane failure?
Targeted testing: swapping the drive between different slots and different controllers. If the error follows the drive, it's the drive. If the error follows the slot, it's the backplane or connector. RAID controller logs: CRC errors on the interface point to the backplane; SMART errors on the media point to the drive.
Do you repair backplanes by resoldering individual pins?
In specific cases yes, where the fault is localised (a bent but unbroken pin, a connector with oxidation that can still be recovered). For most backplane faults, replacement is the more reliable option.
On an NVMe backplane (U.2 with MCIO/SlimSAS), does recabling make a difference?
Yes — cable quality and routing matter a great deal at NVMe speeds (PCIe Gen4/Gen5). A degraded MCIO cable, or one bent outside spec, can produce PCIe errors at a very low rate that still add up to a real impact on latency and throughput. On dense NVMe systems we recable with care.