that would worry me as a concept. As long as the machine is completely shut down then, yes, ought to be OK but even getting BIOS to poll the SATA ports then find the live one and boot from it seems, perhaps I am just supersticious, like it's usng the hardware in a way it's not configured to be used. I can't point to a logical problem with the approach so you can dismiss this but, personally, I would use something like GRUB or even just the BIOS and choose the boot in software rather than powering attached devices off and on.

273's caution seems to me to be warranted. Now I refer to myself as "An Eternal Novice" when it comes to computers, but I honestly believe that the power-only switch can be operated safely: I never turn the box on until I have checked the buttons on the switch and made sure that none of them is depressed. Then I decide which drive I want to boot, depress that button, and proceed deliberately.

Now for the news (and I'm kicking myself for not thinking of this long ago): as statistically unlikely as it is, in fact two of my three SSD drives failed.

Proof of concept: I swapped the positions of SSD 1 and SSD2. SSD 2 has the Debian install, and it is running flawlessly. SSD 1 has a hopeless mess that was supposed to be Slackware. I booted into Debian by depressing button 1 on the switch, not button 2. By the same token, booting into SSD 1, where Slack has collapsed, means pressing button 2.

That's right: why did I completely overlook the possibility of SSD failure, and try to blame everything on my (elderly but still fully functional) mobo? I suspect it was due to my conviction that SSDs are the wave of the future. Maybe they are, but they have life-spans that vary, eh?

I'm going to purchase three brand new spinning disc hard drives.

Does anybody want to recommend which brand I should purchase?? TIA!

That is very interesting. I have not started using SSD's yet - one of the reasons being because of the numerous reports that they tend to fail quite suddenly and without warning - unlike HDD's - which most of the time fail gradually (specially if you have smartd keeping an eye on them and warning you in time). The other reason is that most of my setups need the larger capacities more than they need the extra speed.

As to hard-drive brands, I believe avoiding Seagate is a priority - as their failure rates in the last few years have been quite hight - based on frequent online reports and the number of Seagate drives I had to send back in warranty.

While I agree that it would appear that SSDs can simply fail more often that HDDs, which tend to give some warning, I do not see that as a reason to avoid SSDs. If your backup policy and recovery is dependent upon having warning that your storage will fail then your strategy is a very bad one.
I run my systems with the assumption that, at some point one or more components will just fail and, possibly, take my data and/or the whole machine with them. That's not to say that my backup and recovery strategy is perfect but that I don't avoid SSDs in the hope that a spinny thing won't fail as quick.

My post does not state that SSD's fail more often than HDD's. It says that SSD's more often fail without warning, compared with HDD's - at least based on various online reports.

But I agree with your thoughts on the need for appropriate backups.

On the other hand, just as an example of a particular situation - the last few hdd's which failed in my main laptop started to give SMART errors first. I had a good few weeks to arrange for the purchase of a new hdd, to find a suitable day when I could be without the use of the laptop for a few hours - during which I installed the new hdd and transferred the data - all in an orderly fashion and without stress. Had it been the case with a SDD which might have failed without warning in the middle of a busy day full of urgent work at clients - it might have been a completely different story.

So in practice, for me, it can make a significant difference. Of course, it is possible to arrange for a hot standby laptop constantly at the ready, constantly running and constantly keeping data in sync :-)

Perhaps I ought to have typed "simply fail" -- I meant that they just fail without warning more often not that they, in general, fail more often.

Edit: While, in general, it is more likely that with a SMART error one can simply transfer files it isn't guaranteed that the drive will not just fail with the SMART errors being symptoms of a larger problem. It's often said that when one drive starts to fail in a RAID 5 the chances of another failing while the RAID is rebuilding is fairly high, for example.
If your livelihood or similar depends upon a working laptop every day then you need two laptops and, possibly, some spare parts. If you don't actually need the laptop then, yes, a hard drive gradually failing may be more convenient but, as I mentioned, may amount to the same thing.
Also, remember that storage is just one component and a CPU, motherboard or PSU may just fail.

That is certainly possible - but in practice, maybe I'm lucky, but of all the hdd's with SMART errors I've had to replace, the data was still readable without problems.

I think this refers to the situation when the drives in the array are of the same make, model and age - which I always try and avoid. I've never had a failure so far of a second drive during the rebuilding of the array as the result of changing a failed drive.

Agreed - and I do. I still appreciate the extra notice and time to get things done when it is convenient. At least in my situation.

Theoretically correct. In practice though, hdd's are by far the most frequent points of failure. For a well used machine, every 3-4 years, like clockwork, most of them go. I think I've only ever seen once or twice a dead CPU in 20 years. They don't have moving mechanical parts - they don't wear out like hdds. RAM and mobos - a bit more often, but nothing like hdds. I still have a handful of old servers running on desktop mobos which are closing in on the 12 years mark. I don't have any 12 years old hdds left in production though :-)

All I can say is that I've had HDDs fail on me without any SMART errors. My backups don't rely on me having a warning to get new storage and move the data when time permits.

SMART errors are a great resource to provide you insight on your harddrive's health, but even the healthiest-seeming person can still die instantly from an aneurysm. If you get warnings that your drive might be dying, that's great, but it isn't always going to happen.

1 members found this post helpful.

With you identifying that this is a drive issue, please identify the models of the SSDs that failed. There have been some bad models out there, and it's handy to hear first hand accounts. And update your thread title.

Agreed. That's what backups are for. And RAID arrays - if possible (i.e. - not a regular laptop).

Sorry to everybody for going on a slight tangent in this thread :-)

How recently did you upgrade to kernel 4.4.29? Was it the official Slackware package?
Did you run "lilo", etc after upgrading the kernel?

It may be worth backing out that kernel and either temporarily going back to the official 4.4.14 which comes with Slackware 14.2 (but vulnerable to the "Dirty Cow" bug), or going to 4.4.30.