#funnily enough this is post written from a dell laptop
Explore tagged Tumblr posts
just-watch-and-calculate · 13 days ago
Text
a failing hard drive (and a mental health revelation)
(written by an adhder off their meds: I will go off on tangents)
(the data seems to be safe now)
(this post is about two things which feels a bit weird but the point I want to make is somewhere where both things overlap)
a lesson I learned from this:
back up your data! do it now! yes, NOW!
it started a few weeks ago when I changed linux distro from suse to arch (though it also started earlier. human starts are hard to define. me and my neurodivergence and my pc and hard drive and my mental health struggles that most if not all nd people have in some form didn't just pop into existence when I started installing arch).
I had unplugged two of my drives to protect the data on them from any mess-ups during the install process. (I should've made backups, yesterday evening would have been less scary that way).
the installation went fine, but it took a few days due to mild lack of sleep and me not using archinstall, but I succeded. (and after accidentally uninstalling the login manager on my laptop and briefly reprioritizing to fixing that because I couldn't get the gui to launch manually without the login manager)
and I was proud of my setup. one minor thing, the drive my /home folder (that's where the user accounts' data is) started its off-on spiel it sometimes did when the cable wasn't plugged in properly. (at that point it had not seemed alarming. I had unplugged it prior and was already familiar with this happening)
it had messed up something with i3, I rebooted, it was fine. I knew what drive it was because I can hear its whirr and the clacking of the head going into its off position when it turned off. I love how it sounds. that I can hear how it is 'alive'.
so I opened my pc, unplugged the drive and plugged it back in. and it was fine for a few days.
after it had happened (or maybe only after the 2nd time it happened on arch) I also looked up if there were tools to assess drive health and installed the smartmontools package. I do not know if I would have been a bit slower in noticing when my drive started to fail, maybe the sounds alone would have worried me enough. quite possibly not. ultimately the s.m.a.r.t. output got me to start a backup of the raw data though, but not by actually warning me directly of imminent failure. it estimated about 20000 more hours of use (edit: that is untrue. I just misinterpreted something).
I think it also went off-then-immediately-back-on a few more times before and after that; but that and the following were the two instances that caused my OS tho struggle before the hard drive failed.
after the second time it happened, upon rebooting, the drive would just do its off/on routine while fstab was trying to mount it on 2 consecutive reboot attempts but failed. mildly panicked I booted the live arch usb I had previously installed from and tried to figure out what to do. I didn't back up my data then. I should have. I had planned to set up backup automation then but not yet got around to doing so. I should have done so.
it was unlikely to be the cables' fault, but yet again I unplugged the drive and plugged it back in. yesterday I found out why that was futile. it did its off/on routine after I'd tried getting a more taxing game to run, but i3 was fine, and so I continued, mildly worried. the second time it happened, not long after, I checked the s.m.a.r.t. data. Because I'd sent all the interesting bits of the data to someone the day prior, I was able to look up how it had been before. still, it estimated about 20000 more hours of drive use (edit: that is untrue. I just misinterpreted something). but the count of PhyRdy->PyNRdy was off the charts. from about 300 in a prior power cycle it rose to above 18000. then, eventually it even reached 19000. online advice on this was: back up your data. now.
and so I did, finally, back up my data. the easiest option with now ramping up andrenaline (and a cold I was still recovering from) was using dd. I made sure the infile and outfile were the right way around and off I went, copying all the disk contents to another disk. (or rather, the contents of the partition I was mainly using)
it went ok at first. then the occasional off/on, going by how the drive sounded. ok, I hope you make it, buddy. please, at least long enough to save my data.
then it wasn't just clack-spin-down-fully-then-spin-up-again. it was stumbles, the clack sounded different, had different timing, it didnt spin down fully before spinning up again. (for clarity: by spin down I refer to the motor inside reducing in speed until it is off, like when shutting down my pc.)
the copying stumbled. a bit past halfway it stopped. i/o error. disk dump (dd) had exited. smartctl just told me "inqury failed". my os seemed to think the disk was still there. it just wasn't responding. not presumably unmounted this time, just not responding. the last temperature readout had been a bit above 40°C. high, but it should be fine, right? the PhyRdy -> PhyNRdy statistic was 19806 at last readout. my hard drive was definetely failing now. dying.
I called someone with more IT experience than me; adrenaline was definitely kicking in now. we opted to shut down my pc and try getting the rest of the data after waiting 20 minutes. giving my hard drive a break. it did not make its usual sound when shutting down. it must've spun down at some point before and not spun up again. I braced for the worst, for the first half of the dd to only contain the data I've had older backups of and for the rest to be lost.
my hard drive had just needed some rest. ddrescure came to the rescue, I copied the remainder of the data with not many auditory "complaints" by the hard drive. then turned off my pc to unplug that drive. it could rest now. it had done its part. I no longer needed to fear that my data would be lost. as of writing this I'm combining both halves of the backup on it into one file to mount it and see if my data is intact. odds are looking good but I should make sure. (as of my 2nd draft I'm mounting the file. fingers crossed.) (as of finishing the 2nd draft of this fsck reports the backup's filesystem as having a bunch of errors. I hope my data is salvageable. mounting the file worked but that is a lot of errors from fsck. gonna do a full ddrescue now.) (the ddrescue is nearly finished; the two halves should be recombinable now, ddrescue just has important syntax differences compared to dd. my data is most likely safe.) (the data backup seems to be fine now!)
I miss the way that hard drive sounds. I miss hearing it start. I miss its hum. I miss the hard drive that, quite literally, made my pc home. I only was at home on arch once I had moved my user data to the new home directory. (finally it wasn't my deadname anymore. being trans was now, accidentally, a way to keep my suse and arch user accounts nice and separate)
/home was home, and now it was failing. dying.
how odd, to be feeling (slight) grief for a hard drive.
and now for the mental health realization:
I've anthropomorphized that hard drive in some places in this post. while it was actually failing I was somewhat doing that most of the time (you can make it, buddy! please, please just try to keep going a bit longer. once my data is safe you can get a break.)
and I realized why that is.
one part of it is just that I have the tendency to anthropomorphize things, like the chocolate snowman who I found out had a name only after I had already eaten it. I felt sad, and a bit silly. I'm never buying chocolate with a name like sammy again.
but there's more to it; and it isn't coincidence that I am putting this in a post about a failing hard drive, rather than that chocolate snowman. not only did I humanize that hard drive when it was "struggling" - I also do it the other way around: taking "schedule maintenance or the equipment will schedule it for you" also as a reminder to take the breaks I need because when I need them they will happen eventually. so I better listen to my body and take breaks when it isn't already a bit too late.
I'm treating myself like the machines I humanize because when treating myself as human what sneaks in is treating myself as someone who should be this good little neurotypical girl. treating myself as who I learned I should be instead of treating myself as me.
treating myself as human has failed somewhat. between being trans and being ace and being neurodivergent, I'm not human in the way I was taught I should be. and with this baggage, I need a better angle at this to be able to take care of myself in a way that will allow me to heal.
I stopped trying to view self-care through my neurodivergent mask. it's not this vague lifestyle thing. it is the basis of my well-being.
I started viewing it as 'self-maintenance'. maintenance is important. it is basis for continued operability. that much is simple. it is specific to what is being maintained. I couldn't exactly have treated my hard drive as a dripping faucet. it would have been nonsense for me to try and save my data from my windows installation which I'm much less experienced with. (and besides, it's windows. I actually use OS as an analogy for my neurodivergence sometimes, I do think it works in many aspects. there are many different linux distros. mine takes a bunch of extra work but works better for me.)
it would be nonsense to try treating myself like a neurotypical person. because I'm not. and it would be nonsense to view this "self-maintenance" as something to do only whenever I've got time between my failing hard drive and uni and having gotten sick. it is important.
short bonus bit - the magnus archives has helped me cope with the fear I felt for my data a bit. and since it continues being an intense interest of mine I'd like to discuss that. just beware of spoilers, I'm not sure off the top of my head when the concepts I'm talking about below stop being spoilers
it is interesting that one of the hypotheses of what is going on with the failing hard drive is that the temperature is somehow why that drive is failing - considering the lightless flame is all about things like this data loss. so hi desolation, I hope to never meet you again. please do not cause the digital equvialent of a housefire (overheating(?) /home) ever again.
running the full ddrescue now, it seems putting a cold pack next to it calmed it down. poor thing really was suffering from heatstroke at normal operating temperatures it seems
45 notes · View notes