 So to satisfy some of our own curiosity of the resilience of the ZFS file system in Freenas and specifically this is a craptastic Freenas 11 build on a RAID Z1 with some old hard drives and only 4 gigs of RAM So we're below the minimum RAM that way we could load it up Now we dumped a bunch of data this and then we started power cycling it. Well power cycling it Yeah, it doesn't corrupt it. So we moved on to Popping the memory out live and now I've actually accidentally ripped the heat sink off the memory So it's running really hot and touching components while it's live and it's it's still work So we have randomly been ejecting memory for a little while on this thing and it recovers every single time now A little bit of background lots of concern and I've it's been years I'm not saying you can't corrupt a ZFS file system Of course any file system can fail anything can go completely wrong But we've been trying to simulate something going wrong and it's proved a really daunting challenge here So besides just random power offs while we're writing to it and seeing where those files leave off Which of course they do leave off partially We've also found that if we copy over let's say half a gig file and we Power it off the moment that it says it's corrupt that it says it's finished copying that sometimes that file gets corrupted Not every time though. Sometimes the file copies all the way. It's a matter of how many milliseconds after We have it copied that we eject the memory off it and yes we just been ripping the memory right out of it while it's turned on because I figure that's an even better test because it keeps the CPU doing whatever the CPU is doing and Actually stops some type of the memory from happening So you know pushes some of the caching into memory while it's doing some of the writing to the disk or some type of Processing that's happening while it's writing it across the redis now This is set up as a raid Z1 like I said That's not a lot of redundancy and that's all on purpose for using some old Samsung 160 gig drives we had just laying around and Let's get after a lot of playing with it It's recovered every single time it turns off every time I we've popped different pieces the first memory the second set of memory We did find out if we take out the first piece of memory like this A lot of times. Oh didn't do it this time. It'll actually power off the computer This time it didn't but every time as you can see with the background on the screen over here The system becomes unresponsive. So we turn it off We put the memory back in this is our really heavy-duty testing procedure here memory back in and Power it back on we said it we got sick of you turning it on each time with a USB drive But we were shoving in the power pins So we set it to auto recover on power loss So we just power cycle it each time and every time it's come up We've also been kind of surprised it hasn't destroyed something else on the board So as a last ditched effort here to wreck this thing and we have a feeling this will wreck it as we might introduce some Random voltages around the board while it's writing and see if it corrupts it if it destroys the board at the same time We're curious if we pop these into another machine that we can re-import the radar a but I'm feeling pretty confident That we'll be able to do that because the system is really like I said the resilience of ZFS Like I said, it's not that you can't corrupt a ZFS, but as much effort as we're putting into this I can't believe it's not been corrupted yet. I mean we've we have been power cycling this thing. We've been ejected memory We've randomly unplugged the hard drive as well as writing a plug-in back game But that's a pretty weak test because they're rate Z, you know, even I'm plugging all of them It just stops whatever those files are on there So the final test that we want to do before we declare this is we're going to zap them other board with voltage Randomly on the board if it comes back on great, we'll do it a few times I have a feeling it may not come back on when we do this and if that's the case then We're gonna take the drives and import them in another board the same setup just transplant it over to another board and See what happens. See if the Drives all import perfectly fine, and I'm hardly worried about sacrificing these 160 gig drives In this in this test case scenario that we're trying to do specifically when we do the zapping We're gonna focus on the memory modules that way we corrupt what's in memory with the concept of if you corrupt something in memory It should write the drives wrong and crash all the disks So that's gonna be our next test we're gonna do right now. All right, so we got the free NAS box move your head a little bit See It's still running. It's up on that screen over there So yes, it's on the screen Steve's got science gloves and science goggles and science goggles safety because now We're going to zap the machine. So let's do a quick zap test science We're gonna science this motherboard and we're thinking let's focus on the memory Steve because we want to corrupt the memory while it's writing Hold on Hold on. I got this We're gonna pass it all the way through Are we good? You ready? No, no go Hold on. Let's get the camera close for this. I shouldn't be nervous, but I got safety goggles. We good All right, let's power cycle this Let me set this on I gotta we unplugged for safety reasons We unplugged the ssd on here because we don't want to accidentally destroy an ssd Because we're having a boot off of that But oddly enough I'm plugging the ssd didn't stop it from running. That's the boot device Why didn't we have a boot off of flash drive? Yeah, we just didn't we installed tssd because it was convenient Oh, well, I think that's it for the motherboard. Give me that I'm gonna try and fire it up again That's it. We killed the board. Let's take this piece of memory out and power cycle it Was this the last science? What oh just a flicker try and turn it around again So it did take high voltage to the memory that we've destroyed the motherboard So step two re-import these drives on a different motherboard because I think this is We've yeah, this is fried So running free dance box writing hard drives doing things Um, and I don't know how many volts that is. I don't know. I I have to look it up But it's a lot of also have to make a nice arc and it we arched all the way across the board Then we have we have fried all the logic chips High voltage failure. All right, step b plug in other board and import free nas Well that did it. Um this board is uh won't fried we swapped it over to the other board now and pfs, uh, sorry free nas I want to say pf sensor some reason the other free bsd thing I like is running here Uh booted right up all the files are still there. So even after the While running high voltage zap through that we did But we didn't touch the hard drives that we didn't want to physically destroy them because well, that's pointless But zapping this old motherboard live didn't even corrupt the files and we were writing files to it again Everything up until it decided to shut off from the high voltage is still completely out in there I may do some more follow-ups. I don't know how much more testing we're gonna do We this this is a nicer board and we don't want to wreck this one This is actually a board that's going in for another project But once again, I just in really impressed with free nas. I mean, I do recommend shutting things down properly But all the random turning on and off that we did and we're talking like we've been doing this for a little while now We spent some time just randomly doing it and randomly ejecting the memory and we pulled the memory out so many times Like I said, we even ripped off the heatsink on it. So there you have it free nas running zfs This is free nas 11 and this is a raid z1 configuration Super fault tolerant from random turnoffs and random memory removals and even destruction of the motherboard We just popped it in over here and Brought up the volumes and works great. So thanks for watching if you like the content here like and subscribe And if you had some other thoughts on other ways I should test this We probably could dig up at some point people drop things off for recycling all the time Some other board that we could do this and say hey try this scenario with The free nas system to see if you can, you know, get it corrupted and I'm curious I want to know because I it's hard to fault these things. I know they can fail I've heard people talk about failures they've had with boxes But if you have some ideas of how we can corrupt this in some situation We can come up with let me know in the comments below. Thanks