
On Tue, Dec 15, 2020 at 06:48:32PM +1100, Russell Coker wrote:
How likely is the following error (which happens periodically) to be on the M. 2 SATA device and how likely is it to be on the motherboard?
My guess would be that it's most likely the M.2 SATA device...because, in my experience, drives suck and die a lot - which is why i'll never use less than RAID-1 (or equivalent, such as mirrored drives in ZFS). OTOH, while I've had LOTS of mechanical hard drives die on me over the years, I've only ever had one SSD die (and even that died "gracefully" - could still be read, but writess failed). SSDs are, IME, a lot more reliable than spinning rust. But why rely on a guess when the obvious thing to do is to test it? 1. Try the M.2 device in another machine If you don't have another motherboard with M.2 slots free, you can get reasonably priced PCI-e adaptors that can take anywhere from 1 M.2 drive (using 4 PCI-e lanes) to 4 M.2 drives (using all 16 PCI-e lanes). These are a useful thing to have around, so it wouldn't be a one-use waste of money. 2. Try another M.2 device in the motherboard. The cheapest M.2 drive available now is around $30 for 120GB. e.g. $ cplgrep -p m.2 | head -n1 32 Kingston SA400M8/120G A400 M.2 SSD 120GB You are, IMO, better off just buying another M.2 the same size or larger (if it turns out to be the drive that's failing, you can immediately use it as a replacement. Otherwise, you've got a spare, or a drive to use in another machine). BTW, if your motherboard supports it, get M.2 NVME rather than M.2 SATA - there's very little difference in price, and the NVME will be around 4 to 6 times faster - depending on brand and model, from ~2500 MB/s up to ~3500 MB/s for PCI-e 3.0 NVME vs ~550 MB/s for SATA. For PCI-e 4.0 NVME, it could theoretically get up to nearly 8 GB/s (less protocol overhead), but current models max out around 5.5 or 6 GB/s. PCI-e 5.0 will double that again in a year or three if SSD speeds keep up with PCI-e bus speeds. craig -- craig sanders <cas@taz.net.au>