HAVE YOU BACKED UP YOUR HARD DRIVE LATELY

[Diana]

So a couple weeks ago my PC's hard drive up and DIED and guess who hasn't backed up their hard drive since like bumblefuck who knows when and lost EVERYTHING. >:[ Photos, roleplay content, iwaku shit. Everything. I had to wait a couple weeks for a new laptop. Now I am having to start from scratch.

It's not the end of the world and nothing that'll fuck me over was lost BUT IT SUPER SUCKS SUPER HARD.

I'm a bit discombobulated now and trapped in "how do" mode, cause I'm used to laptop being PLAYTIME and PC being WORKTIME so my brain is trying to form new habits and get used to new programs and whatnot. D:

And yes, yes I do now have an external portable hard drive for backups now. >_>

HAVE YOU EVER LOT YOUR HARD DRIVES AND THEN CRIED FOR A WEEK?

 

[Lyrikai]

TWICE in the past I had two separate computers suddenly up and die on me when I didn't have backups on all my stuff and I lost eeeeeeverything.
Games, music, RPs, stories I wrote, photos that can never be recreated. I was absolutely devastated.

Now I backup everything in paranoia between an external hard drive, some SD cards, and a lot of flash drives. I have copies of my copies! I will never lose anything ever again!!!!!!!!

 

[tsukasa]

I run Linux (no, I don't touch Windows or Apple stuff; they suck) and I store my data on an NFS server in my house so I have centralised access to all of my files which is running RAID 1 (I plan to increase it to RAID 5/6 in the future), meaning the loss of a disk (or multiple disks) means the mirrored data isn't lost since it's copied verbatim across disks when files are copied to only one of them. I have a scheduled backup to a different RAID array multiple times per day to mitigate loss of files due to corruption or accidental deletion, too, as well as using Bcachefs filesystem with CRC64 checksums to mitigate corruption of files when something hangs or crashes, or there's a loss of power to the systems. Everything is copied off-site (encrypted, authenticated, and integrity-checked, of course) to a cloud service after the scheduled backups have been completed to mitigate loss of my entire server or house. This gives me 4 layers of protection against data loss.

As for my phone, I use Syncthing to back up my home directories across all of my Android user profiles to the NFS server (giving my phone data the same protection as my PCs), while using GrapheneOS' Seedvault to back up the app data (Seedvault is a little broken at the moment, but they're working on a new backup solution).

 

[Astoria]

I DO keep backups but not regularly because I forget. Though I actually did run one a couple weeks ago, so go me!

I've got an external HDD that can just about fit two full images of my SSD, but I'm due for an upgrade to a better backup solution soon. I think I've only lost data once a long time ago when I was a dumbass and formatted the wrong drive. Thankfully I got most of it with recovery software and then I started doing backups.

If the lost data was super important, there's always paying out the nose at a data recovery service...

 

[ImaginationGoneWild]

I write everything down on paper in case my laptop dies. Or transfer stuff on usb sticks.

 

[unanun]

@Diana , if most of your content is text based, the free tiers of OneDrive, Dropbox, or Google Drive will easily hold your materials (about 5 GB for each). If you have amazon prime, you have unlimited storage with them I think. You set up a folder that automatically syncs, and you can rest worry free. Also, if your new laptop has a SSD (solid state drive) then it's basically invincible, so hopefully you won't have any more problems from now on.

 

[tsukasa]

Also, if your new laptop has a SSD (solid state drive) then it's basically invincible, so hopefully you won't have any more problems from now on.

SSDs are more likely to fail than HDDs if large amounts of data is written to them, since the cells have a certain amount of wear they can take (measured in terabytes written). This occurred to me, recently. Tools can be used to view the estimated remaining life (at least on NVMe SSDs). HDDs, while slower and using older technology, have the advantage of writes not wearing them down, only the power-on time and mechanical failure.

 

[Havoccultist]

I've got a USB thumb drive I backup my important IRL/college related documents to, but that's about it. The rest of the stuff on my current laptop isn't actually that important. Settings can be changed from factory since I do use Windows (no reason not to for me as I genuinely have no real reason to use a distribution of Linux personally beyond learning it for eventual work I guess), my Steam library can be reinstalled, my images could be transferred or-barring that-redownloaded entirely, bookmarks can be exported and imported assuming Brave doesn't just save them across installations, and the few additional programs I have like OBS, Wireshark, PyCharm, Malwarebytes, etc., can also be reinstalled on a new machine.

So hard drive or SSD failure isn't all that terrifying of a concept for me.

SSDs are more likely to fail than HDDs if large amounts of data is written to them, since the cells have a certain amount of wear they can take (measured in terabytes written). This occurred to me, recently. Tools can be used to view the estimated remaining life (at least on NVMe SSDs). HDDs, while slower and using older technology, have the advantage of writes not wearing them down, only the power-on time and mechanical failure.

While true, this would take large amounts of writes. So much so that the user's machine is likely to die long before it reaches that point due to planned obsolescence and most laptops (at least since that's my current point of reference rather than a desktop) having a three year lifespan themselves. To say nothing of the general wear and tear that is usually incurred over that timeframe. Now if you're writing and erasing large amounts of data, such as with a game, maybe you'll wear it down faster. But again, you can always move on to a new machine thanks to cloud storage and things like Steam libraries being contiguous across installations (at least in regards to what games you own and can re-install). But yeah, with all that in mind, I'm not too worried if my SSD shits itself. I've got what I need from it, and can always send the machine back to Falcon Northwest for repairs/a replacement.

Citation: SSD Lifespan | How long does an SSD last? - IONOS

 

[tsukasa]

While true, this would take large amounts of writes. So much so that the user's machine is likely to die long before it reaches that point. Citation: SSD Lifespan | How long does an SSD last? - IONOS

I wouldn't say "long before a user's machine dies"; it's completely dependent on how much you write to it. The average terabytes written is around 400-700 from my experience, but if you're willing to pay a little more (SSDs are cheap nowadays), you can get upwards of 1000 terabytes written. My SSD died after 3 years (I/O errors and dropping to a read-only state) and I was writing tens of gigabytes to it on an almost-daily basis, so it shouldn't be concern to most people who aren't enthusiasts like myself, but it's still true that SSDs aren't "invincible" compared to HDDs (although they are better in almost every other way, excluding the wear and security of erasure). A simple mitigation is just to keep a backup of any storage media, regardless of HDD, SSD, flash, whatever; there is no substitute for a complete copy in case a failure (no matter how) does occur. As the saying goes, "If it doesn't exist in 3 places, it doesn't exist at all"; one of those places should be off-site.

 

[Havoccultist]

While true, this would take large amounts of writes. So much so that the user's machine is likely to die long before it reaches that point. Citation: SSD Lifespan | How long does an SSD last? - IONOS

I wouldn't say "long before a user's machine dies"; it's completely dependent on how much you write to it. The average terabytes written is around 400-700 from my experience, but if you're willing to pay a little more (SSDs are cheap nowadays), you can get upwards of 1000 terabytes written. My SSD died after 3 years (I/O errors and dropping to a read-only state) and I was writing terabytes to it on an almost-daily basis, so it shouldn't be concern to most people who aren't enthusiasts like myself, but it's still true that SSDs aren't "invincible" compared to HDDs (although they are better in almost every other way, excluding the wear and security of erasure). A simple mitigation is just to keep a backup of any storage media, regardless of HDD, SSD, flash, whatever; there is no substitute for a complete copy in case a failure (no matter how) does occur. As the saying goes, "If it doesn't exist in 3 places, it doesn't exist at all"; one of those places should be off-site.

While true, I think it's safe to assume that most are not enthusiasts nor writing several hundred terabytes a day/month. Several gigabytes if they're gamers maybe, but not nearly that much. I'm not saying that you shouldn't have backups, but the average user is likely to see another failure of their machine's hardware or software long before the SSD or HDD goes. At least if we're talking average use and not using a personal laptop as a server or the like.

And this isn't accounting for how well said machine was built naturally, because if it were then I'd recommend avoiding HP like a plague.

As far as mitigation goes, for me personally if it isn't kept in some physical form somewhere then it's pretty much fucked. Either human error or a software or hardware failure can take it out, to say nothing of outages or bad actors if it's on the cloud or a database somewhere, and if you forgot to backup or worse that backup fails, it's gone. It's why I keep all my important documents in one place physically, and minimize the amount of important info I need to keep on any one machine or device. At least if documents get misfiled you can always correct that. Sure it won't help if, say, your storage area catches on fire, but nothing's that foolproof. Not even black holes if they evaporate, and those would arguably be the best storage mediums out there.

Still, I'd prefer a physical backup exist rather than purely digital ones, or ones that are written to a disk or drive.

 

[tsukasa]

...due to planned obsolescence and most laptops (at least since that's my current point of reference rather than a desktop) having a three year lifespan themselves.

This is actually the worst part of technology; companies making products end-of-life while they are still perfectly usable without any organic issues. As a security researcher, I cringe so hard when I see people using stuff which is EoL because "it still works", despite them no longer receiving security updates. The particular OS I use on my phone refuses to support devices for more than 1 year after EoL because there is no point (and even that 1 year is classified as "harm reduction" since the proprietary patches stop and the OEM no longer releases them, making it impossible to fix security issues in the firmware/drivers), while most other projects (example) think it's simply enough to update what they can and leave the important firmware/drivers out, which puts their users in danger by not being honest about what they can provide...

This is a major reason Google has switched away from Qualcomm, who refuses to support their SoCs for more than 4 years (used to be 3 years), to their own SoC in order to provide 5 years of support, which was then increased to 7 years this year. There needs to be laws to force OEMs to provide updates for a minimum amount of time, and no less than 4-5 years.

 

[Havoccultist]

...due to planned obsolescence and most laptops (at least since that's my current point of reference rather than a desktop) having a three year lifespan themselves.

This is actually the worst part of technology; companies making products end-of-life while they are still perfectly usable without any organic issues. As a security researcher, I cringe so hard when I see people using stuff which is EoL because "it still works", despite them no longer receiving security updates. The particular OS I use on my phone refuses to support devices for more than 1 year after EoL because there is no point (and even that 1 year is classified as "harm reduction" since the proprietary patches stop and the OEM no longer releases them, making it impossible to fix security issues in the firmware/drivers), while most other projects (example) think it's simply enough to update what they can and leave the important firmware/drivers out, which puts their users in danger by not being honest about what they can provide...

This is a major reason Google has switched away from Qualcomm, who refuses to support their SoCs for more than 4 years (used to be 3 years), to their own SoC in order to provide 5 years of support, which was then increased to 7 years this year. There needs to be laws to force OEMs to provide updates for a minumum amount of time, and no less than 4-5 years.

I know there's at least an EU ruling that's forcing manufacturers to support devices even after the warranty goes as part of a measure to combat planned obsolescence, but that's unlikely to happen in the states. We seem to have an allergy for sensible checks and balances on the tech sector.

 

[tsukasa]

I know there's at least an EU ruling that's forcing manufacturers to support devices even after the warranty goes as part of a measure to combat planned obsolescence, but that's unlikely to happen in the states. We seem to have an allergy for sensible checks and balances on the tech sector.

I think Germany has already passed a law to force OEMs to support smartphones for a minimum of 5 years, but I'm unsure how this will turn out. If that's true, full support from me; I'm tired of seeing people using insecure stuff because the devices work fine other than updates they need but don't think they do. It really is bold to blame users for being hacked etc when companies are encouraging using blatantly insecure devices by dropping support while they work perfectly fine.

 

[Havoccultist]

I know there's at least an EU ruling that's forcing manufacturers to support devices even after the warranty goes as part of a measure to combat planned obsolescence, but that's unlikely to happen in the states. We seem to have an allergy for sensible checks and balances on the tech sector.

I think Germany has already passed a law to force OEMs to support smartphones for a minimum of 5 years, but I'm unsure how this will turn out. If that's true, full support from me; I'm tired of seeing people using insecure stuff because the devices work fine other than updates they need but don't think they do. It really is bold to blame users for being hacked etc when companies are encouraging using blatantly insecure devices by dropping support while they work perfectly fine.

Citations: European Parliament takes on greenwashing and early obsolescence to empower consumers.

EU law will ensure product durability.

Brussels targets greenwashing, planned obsolescence in new EU consumer rules.

It's also ballsy to blame users when half the hacker groups out there go the data breach route and target the source, the company's servers, or employees on the same network or who have access to them, rather than the users themselves. Like... what was the user supposed to do at that point? That's on the company for having a hole in their security, not the user.

And that's not even getting into the whole third-party thing, or user data being sold almost constantly.

 

[Astoria]

While true, this would take large amounts of writes. So much so that the user's machine is likely to die long before it reaches that point. Citation: SSD Lifespan | How long does an SSD last? - IONOS

I wouldn't say "long before a user's machine dies"; it's completely dependent on how much you write to it. The average terabytes written is around 400-700 from my experience, but if you're willing to pay a little more (SSDs are cheap nowadays), you can get upwards of 1000 terabytes written. My SSD died after 3 years (I/O errors and dropping to a read-only state) and I was writing tens of gigabytes to it on an almost-daily basis, so it shouldn't be concern to most people who aren't enthusiasts like myself, but it's still true that SSDs aren't "invincible" compared to HDDs (although they are better in almost every other way, excluding the wear and security of erasure). A simple mitigation is just to keep a backup of any storage media, regardless of HDD, SSD, flash, whatever; there is no substitute for a complete copy in case a failure (no matter how) does occur. As the saying goes, "If it doesn't exist in 3 places, it doesn't exist at all"; one of those places should be off-site.

Oh yeah, I lived life dangerously with a single factory refurbished 1TB SSD for seven years. I was redlining that bugger for most of those seven years at 95% capacity with no budge to the 100% health rating with the number of read/writes I was doing. Somehow got away with it.

Meanwhile my friend maxes out his case with drives and yet he somehow he managed 307TB writes on his relatively new OS drive with SMART throwing errors everywhere at the very end.

 

[tsukasa]

Oh yeah, I lived life dangerously with a single factory refurbished 1TB SSD for seven years. I was redlining that bugger for most of those seven years at 95% capacity with no budge to the 100% health rating with the number of read/writes I was doing. Somehow got away with it.

Meanwhile my friend maxes out his case with drives and yet he somehow he managed 307TB writes on his relatively new OS drive with SMART throwing errors everywhere at the very end.

I currently have 5 SSDs in one of my servers which are in a RAID 1 configuration and I will probably require replacing them at some point within the next year with how much I write to them (they are brand new, only weeks old at most). With Bcachefs, I can use the writeback caching feature which allows copying in-use data from another disk and caching it on another one; I'm using HDDs for cold storage and copying that data to the SSDs when it's called; this allows me to have the performance of the SSDs, while having the storage of the HDDs, but it also allows me yet another layer of protection since failure of all SSDs (unlikely, but more layers is never a bad thing if your data matters to you) gives me the HDDs as a final fallback. The data is written to the fast drives, and copied back to the HDDs in the background to prevent a performance reduction.

Bcachefs also allows me to transparently compress all data in memory before writing it to the drives, making it faster to write to HDDs due to smaller file sizes, and reducing wear on SSDs due to having less to write to them.

 

[unanun]

My guess for Diana's use case is that a SSD would be far more reliable than a HDD, generally, but I agree that SSDs do not survive thrashing. Hopefully she's not caching a write-heavy SQL database on there :P

 

[tsukasa]

My guess for Diana's use case is that a SSD would be far more reliable than a HDD, generally, but I agree that SSDs do not survive thrashing. Hopefully she's not caching a write-heavy SQL database on there :P

Databases often run in memory. I run PostgreSQL for a few services I run, and it's better to run those on a server which is left on rather than shut down; you flush the in-memory data to disk when you need to shut the server down for updates or whatever reason, and a UPS should be used to give you time to flush the data in the event of a power cut.

 

[Astoria]

My guess for Diana's use case is that a SSD would be far more reliable than a HDD, generally, but I agree that SSDs do not survive thrashing. Hopefully she's not caching a write-heavy SQL database on there :P

Databases often run in memory. I run PostgreSQL for a few services I run, and it's better to run those on a server which is left on rather than shut down; you flush the in-memory data to disk when you need to shut the server down for updates or whatever reason, and a UPS should be used to give you time to flush the data in the event of a power cut.

Funny personal story about UPSs, that. Prior to me getting one, the last time we had a blackout or severe power event was the Big One in 2003. Even knowing how clean the energy is in my area aside from the very rare dip or surge, I got a nice big UPS for my setup because I like my electronics.

Well y'know what? The one week where the UPS was down because the batteries needed replacing? The first blackout in almost 20 years. Knocked out my equipment. Not permanently, thankfully, but it's a hilarious story in hindsight...

 

[tsukasa]

Funny personal story about UPSs, that. Prior to me getting one, the last time we had a blackout or severe power event was the Big One in 2003. Even knowing how clean the energy is in my area aside from the very rare dip or surge, I got a nice big UPS for my setup because I like my electronics.

Well y'know what? The one week where the UPS was down because the batteries needed replacing? The first blackout in almost 20 years. Knocked out my equipment. Not permanently, thankfully, but it's a hilarious story in hindsight...

I once deleted the current kernel from my system (running Linux) before moving the newly-compiled one to the /boot/ directory and configuring the UEFI to boot it, because I thought nothing would happen just that once, knowing I should have moved it before deleting the old one. I accidentally rebooted the system after deleting the old kernel and before moving the new one, so my system couldn't boot...

It was an easy fix with a live CD I have on a flash drive; just boot into that and mount the OS partition (while running the live CD's kernel), then move the new kernel over to /boot/. I don't delete old kernels before I move the new ones, now...