Is the Linux kernel more sensitive to HD failure?

Roken · 2013-04-17 20:22:12

I ask because of behaviour I seem to have observed. I have three HDs (spinners) and an SSD in my system. One of the HDs houses Windows 7, which I rarely boot, but keep around just in case. One HD is my primary data drive, and one was my original Linux drive before I got the SSD. Arch has been migrated to the SSD and happily resides there.

About 3 months or so ago, the Linux kernel started complaining about ata errors on one of my HDs (the windows one), thoug SMART reported no errors at the time, and neither did Windows. 3 - 4 weeks on I got SMART reporting errors, and eventually, even Windows complained. I replaced the HD and migrated everythign with Clonezilla, and now it's as happy as the proverbial Larry.

In the past few days, my Data drive started rare clicking, and at teh same time, the Linux kernel started complaining about ATA errors. SMART reports clean, and Windows says nothing. However, the clicking itself is warning to me that I should start to look to replace and clone it.

What strikes me is that, apart from my own ears, Linux has complained a long long time before Windows, and even ahead of SMART. So, is the Linux kernel inherently more sensitive to HD failure?

alphaniner · 2013-04-17 21:02:27

[contrarian_hat]Or maybe just more noisy. I'm guessing the kernel made its complaints known by spewing them onto the command line. In fairness, there's not really an equivalent for that in Windows. I bet the event log would have some record of it, though.[/c_h]

Last edited by alphaniner (2013-04-17 21:03:38)

cfr · 2013-04-18 01:13:02

Do you mean smartctl showed nothing? Or just that the "overall health checked" was passed? In my rather limited experience, those are very different questions.

Roken · 2013-04-18 06:51:00

@ alphaniner. It's possible that Windows logs would have more, though eventually Windows did start popping up warning messages of imminent doom

@cfr. I mean smartctl showed nothing, and tests were passed when run, initiallly. Only later did SMART errors start to appear, which struck me as odd. I would have thought SMART errors would occur first

brebs · 2013-04-18 07:20:14

Roken wrote:

rare clicking

There's the infamous hdparm -B <value> - have you checked what value Linux is using, vs Windows?

alphaniner · 2013-04-18 14:01:24

My point is just there are a lot of "bad" ata events and it's not unreasonable if Windows isn't configured with pop-ups for all of them. If the events don't show up in the event log, that's another matter.

If both OSs are intact, you could probably use the systemd journal and the Windows event log to get a more nuanced idea of what went down.

Roken · 2013-04-18 16:19:37

@ brebs APM is off, so can't think that's the problem

I may look into the logs more deeply, but historically I've always treated a clicking drive as a dying drive, and since this one is about the same age as the previously dead drive (same make/model, too) it's not unreasonable to assume it's at end of life.

On the plus side, it's only 500Gb and I can replace for £28 - so it's not going to kill me.

Last edited by Roken (2013-04-18 16:23:50)

Arch Linux

#1 2013-04-17 20:22:12

Is the Linux kernel more sensitive to HD failure?

#2 2013-04-17 21:02:27

Re: Is the Linux kernel more sensitive to HD failure?

#3 2013-04-18 01:13:02

Re: Is the Linux kernel more sensitive to HD failure?

#4 2013-04-18 06:51:00

Re: Is the Linux kernel more sensitive to HD failure?

#5 2013-04-18 07:20:14

Re: Is the Linux kernel more sensitive to HD failure?

#6 2013-04-18 14:01:24

Re: Is the Linux kernel more sensitive to HD failure?

#7 2013-04-18 16:19:37

Re: Is the Linux kernel more sensitive to HD failure?

Board footer