It seems like Apple is finally going to remove support for AFP from macOS, twelve years after first moving from AFP to SMB for its default network file-sharing technology. This change shouldn’t impact most people, as it’s highly unlikely you’re using AFP for anything in 2026. Still, there is one small group of people to whom this change has an actual impact: owners of Apple’s Time Capsule devices. Time Capsules only support AFP and SMB1, and with SMB1 being removed from macOS ages ago, and now AFP being on the chopping block as well, macOS 27 would render your Time Capsule more or less unusable.
It’s important to note that the last Time Capsule sold by Apple, the fifth generation, was released in 2013, and the product line as a whole was discontinued in 2018. If you bought a Time Capsule in the twilight years of the line’s availability, I think you have a genuine reason to be perturbed by Apple cutting you off from your product if you upgrade to macOS 27, but at least you have the option of keeping an older version of macOS around so you can keep interacting with your time Capsule. It still feels like a bit of a shitty move though, as those fifth generation models came with up to 3TB of storage, which can still serve as a solid NAS solution.
Thank your lucky stars, then, that open source can, as usual, come to the rescue when proprietary software vendors do what they always do and screw over their customers. Did you know every generation of Time Capsule actually runs NetBSD, and that it’s trivially easy to add support for Samba 4 and SMB3 authentication to your Time Capsule, thereby extending its life expectancy considerably? TimeCapsuleSMB does exactly that.
If the setup completes successfully, your Time Capsule will run its own Samba 4 server, advertise itself over Bonjour (show up automatically in the “Network” folder on macOS), and accept authenticated SMB3 connections from macOS. You should then be able to open Finder, choose Connect to Server, and use a normal SMB URL instead of relying on Apple’s legacy stack. You should also be able to use the disk for Time Machine backups.
↫ TimeCapsuleSMB
It’s compatible with both NetBSD 4 and NetBSD 6-based Time Capsules, although you’ll need to run a single SMB activation command every time a NetBSD 4-based Time Capsule reboots. This will also disable any AFP and SMB1 support, but that is kind of moot since those are exactly the technologies that don’t and won’t work anymore once macOS 27 is released. The installation is also entirely reversible if, for whatever reason, you want to undo the addition of Samba 4.
This whole saga is such an excellent example of why open source software protects users’ rights, by design.
In all fairness, most time capsules are destroyed by groundwater, making the recovery of the contents impossible, so they kind of warned everyone that there is a good chance the contents placed in the Time Capsule will become irrecoverable/inaccessible.
On a more serious note, can’t you just add support for AFP to MacOS 27 via a third-party driver or something?
So, I guess people who bought RTX 20-series GPUs just to see support for Nvidia 3D Vision removed 7 months later have a genuine reason to be super-extra-perturbed?
Imagine having bought multiple 3D Vision monitors (to implement 3D Vision Surround) and also Nvidia’s 3D Vision glasses, and less than 7 months later, you can’t update to a newer driver without losing the ability to use all that hardware you paid for (note: most 3D Vision monitors don’t support HDMI 3D, the stereoscopic mode only works with Nvidia GPUs). As of 2026, using 3D Vision in Windows 10 requires blocking driver updates for the GPU using the Group Policy Editor so that Windows Update won’t helpfully (and without confirmation) update your 7-year old GPU driver.
Or, you know, all the people who bought the Spotify Car Thing only to have it bricked, but at least they got a refund (something Jensen “Leatherman” Huang never did for people who bought his company’s 3D glasses). Then again, imagine and buying a used car without a touchscreen on the assumption you can use Spotify’s Car Thing as a thing for your car.
It’s this kind of stuff that makes me buy software keys from shady “key resellers” with no remorse (in fact, I do it with pride). Or teach people how to subscribe to various services in (devalued) Turkish Liras using a VPN.
Speaking strictly of Microsoft products, that’s not really necessary these days if you can find that huge cemetery that can freely activate your software using official methods provided by Microsoft.
More generally speaking, yeah planned obsolescence is a real bitch. A few years ago I bought a new model of a guitar->computer interface, and less than a year later it went out of support and the license key for the software expired. I couldn’t install its software in macOS or Windows once it expired, even though it was supposed to be a “lifetime” license that was tied to the hardware. It ended up being moot since 1) I have developed a severe case of trigger finger on my left hand that has made me have to stop playing guitar, and 2) I no longer have any macOS or Windows devices I could connect it to. Out of curiosity I plugged it into one of my Linux boxes and it shows up in dmesg as an audio source, but I can’t get it to work with any music software, so even that’s not an option. Thankfully it was only $100 but I only got to actually use it for a few months until the “lifetime” license died. That shit should be illegal.
Honestly this is probably more of a kindness by apple at this point.
These things came out in 2008… But let’s be generous and say it’s the 2018 date. That’s well beyond the life expectancy of any hard drive installed.
If you are relying on a 20(ish) year old 500gb disk to be your sole backup solution in 2026, good luck to you.
I’ve never used a Time Capsule, can the hard drive be changed without bricking the device? I can’t imagine buying a NAS-like device without the ability to swap out storage, but then it’s Apple so it would be par for the course if it’s a sealed unit that has to be physically broken to change drives.
But yeah, these days $200 will get you a two-bay NAS with 2.5GbE, Wi-Fi 6, and USB 4; if you’ve got the drives to chuck into it laying around it’s a no-brainer.
Very Apple, you Can, but not supported.
It’s a semi-sealed unit.
A NAS is the sort of the last piece I need to complete my home lab and security set up, but I’ve been having a hard time picking a lane on that one. I actually have a really old 2U rack mounted QNAP unit that I got from work. But it’s out of support, and I’m sure it’s power hungry – overkill for a homelab for sure. Also, prices for the drives…
Still, I wonder what’s out there. I also need that fiber to hurry up and come to my street for the delicious upload speed advantage. Once I get that, I’ll probably buy something good. Probably will pick up a small ugreen or synology unit. Is there something better? Maybe something I can run open source software on and supports a mix of nvme and spinner drive bays for cheap? I’d probably want at least 4 bays, but 2 might be enough.
I don’t know why I’m sharing this info – but I will say, as much as I generally agree with Thom’s takes on corporate abuse, ending support for a unit that wasn’t actively developed for 13 years, and the last unit was sold 8 years ago – that’s weird. Companies can’t be expected to support old out dated technology forever, and in this case they gave plenty of time before pulling the plug. AND – there’s an easy workaround, which is what the post is about. It’s a little much in this case.
CaptainN-,
Anything with x86 hardware generally works as long as you can throw disks in it…. If it’s ARM based, you may need a custom distro. Sometimes you’ll find others have already put in the work. I had a Buffalo branded NAS and downloaded a fork of debian linux for it, which worked great. But my issue with these is that even if you get it working, they’re non-standard and can end up with repos that are locked in time or worse disappear altogether.
I don’t have much experience with BSDs and I am curious if the situation with non-standard hardware is different than it is with linux?
Is it planned in your area? Here I’m living with the optimum online monopoly. Several years back verizon had a fiber network expansion but they stopped before reaching us with no future plans. It sucks because when one is served by a monopoly it’s customers get overcharged for service that under delivers 🙁
Even though you can buy large drives, I wouldn’t consider less than 4. I would only consider using RAID and RAID-1 halves the capacity. Two disks also can’t benefit from the performance gains of RAID-10 or the improved redundancy of RAID-6. Or if you are into the flexibility of un-RAID, you really need to be able to support more disks to see the most benefit there.
You’re right that’s par for the course….however I would say that when the product was actually being sold as new in 2016, only supporting SMB version 1 was absolutely pathetic. SMB2 came out publicly in 2006 and SMB3 was from 2012. IMHO official support from apple should have been better at least while it was being sold.
A while back I helped a friend set up a UGreen 2-bay NAS that was based on the N100 SoC and ran their in-house Linux-based OS. It was actually a very solid unit, and the web-based interface was like running a full desktop OS in the browser. It was fast and familiar feeling, kind of a mixture of Windows 7 and macOS from the same era on the front end, but with modern server-specific features. It had built in support for Docker containers, which I used to set up his Emby server and the various services to download media.
He quickly outgrew the storage, so he got an AOOSTAR 4-bay NAS with a slightly faster N150 SoC, but unfortunately no NAS software; it shipped with Windows 11 of all things. I set him up with OMV and the same media software as on the old unit, and he’s golden.
All of that to say either of those are solid choices in my book; if you want a good pre-installed NAS software go with UGreen, or if you prefer to roll your own go with the AOOSTAR or a similar unit based on the N150.
Adurbe,
I believe this depends very much on uptime. A drive that’s used an hour or less per week won’t incur the same rate of mechanical degradation as one that’s kept busy. Anecdotal evidence isn’t terribly useful, but I do have HDDs that have lasted decades. Ironically I’ve experienced a lot more failures in the early days of the “bathtub curve”.
Most of the major papers I’ve seen about disk drive longevity study data center applications where drives operate 24×7. Here’s one such excerpt…
https://www.backblaze.com/cloud-storage/resources/hard-drive-test-data
As it relates to our discussion about home NAS, if one got a HDD disk in 2018 and the annual failure rate under normal use is ~1%, the expected failure rate at 8 years would only be 8%. I don’t want to assert that actual failure rates are perfect exponential curves. but regardless the math projects a 50% failure rate.occurring at 69 years.
So that drive from 2018 is about 8% likely to fail, statistically speaking. Risk tolerance not-withstanding, the expectation that the NAS still works isn’t so unjustified.
The drives do come under quite consistent load. Generally hourly incremental and daily full (I think that’s default but haven’t checked) .
Part of the issue is it’s compounded by the hardware choices where many use WD Green drives which are not designed to be NAS/Backup and are designed by the manufacturer for a 3-5 year lifespan.
While you are correct it Might last longer and usage will vary greatly. Even taking your figures, 8% chance of losing all my data (backup) doesn’t fill me with confidence. But my expirence of hard disk failures is that they either fail fast or failure rate accelerated over time and wasn’t as linear as that data suggests.
If it had dual drive, we’d be talking different risk profile but there are certainly people/small businesses out there that have no redundancy other than this.
Again this is focusing on the 2018 model, earlier ones are at higher risk
Adurbe,
Yes, clearly it depends on usage. Some will push it harder than others. Conceivably such daily backups may well be within manufacturer specs when they publish AFR and MTBF, figures so daily backups aren’t necessarily a problem.
They’re typically warranted for a 3-5 year lifespan. meaning the failures have to be kept low enough that the manufacturer doesn’t have to replace a large percentage of them in that period. This doesn’t automatically mean the mechanics suddenly become non-viable after warranty. Of course if manufacturers employ planned obsolescence tactics to make products suddenly less usable, that changes everything. Although I haven’t heard of HDD manufacturers employing such failure acceleration schemes. I personally have several drives that are decades old and haven’t failed. (I use them redundantly so their failure would be ok). Still, if you are aware of any that are designed to fail after warranty, I would appreciate the information because I do like to learn about these things. If you don’t have information about this though, I hope you don’t mind my treating it as an allegation rather than a fact.
I don’t believe physical products follow precise exponential degradation, so the math isn’t exact. Nevertheless, 8% faillure rate after 8 years is actually good. Even if you bought a brand new disk each year, you’re still looking at 1% AFR. Real products can have a bathtub failure curve, you may actually be dramatically increasing your exposure to defective products and putting your data at greater risk with a new HDD than a proven three year old disk with no errors.
I don’t have the data on this bathtub curve for HDDs unfortunately, but just to illustrate the idea I’ll point to infant mortality where there are more deaths in the first year of life than the next 4 combined.
https://web.archive.org/web/20140324141053/http://www.who.int/gho/child_health/mortality/neonatal_infant_text/en/
Statistically speaking, something can have a higher expectation of survival after 5 years than when it’s new.
Admittedly this is all handwaiving, without data specific to HDDs I can’t prove it, but it seems plausible to me that the bathtub curve skews more failures and risk with new hardware and less risk with older hardware.
Of course I’d love to have more data, but I don’t think 8 year old drives are statistically as failure prone as you are implying.
Time machine backs up hourly by default…
A disk that spins up and down repeatedly is under a lot more stress than one which stays running 24/7.
A disk that is subjected to fluctuations in temperature is also under more stress.
Backblaze run their disks in a datacentre, they run 24/7 in a temperature controlled environment. You cannot directly compare that to a home environment where it could well be subjected to multiple start/stop cycles both intentional and otherwise, vibration and movement, fluctuating temperatures, dust, fluctuations in power supply, etc.
bert64,
You’re making too many assumptions here. Not everyone will be backing up every hour of every day. You also assume it will degrade worse than a disk in 24×7 operation, but I don’t know that to be true. In the absence of data, I’m not sure how many times we have to power up and down the disk to match the mechanical stresses of head actuators operating 24×7. I don’t know the answer, but without the data isn’t it fair to say you’re just making an educated guess?
So while I can admit these variables exist, I’m not going to concede the stress on disks in Apple Time Machines is higher than other normal or even server use cases. We really need more data before making factual claims of degradation under specific applications like this.
Say we had data showing that the average failure rate of disks inside of apple time capsule products fails at 1.5% AFR instead of 1% AFR elsewhere, this would clearly show 50% more degradation per year, but do you have data showing anything like this to be true? If not, then can we agree that these are just assumptions? We can speculate all day, but we need data in order to make definitive conclusions. I hope that we can agree on this.
I would guess Asahi Linux can support AFP with a few simple commands?
Don’t need to mess with the Apple Time Capsules, just connect to it with Linux or another OS that the supports AFP protocol.
I don’t understand why it was easier for this guy to fully update the time capsule OS than to simply compile netatalk (opensource AFP) and some samba v1 version on his mac.
How would installing an AFP server and an SMB server on the client system solve connecting to another AFP server and SMB server on the Time Capsule?
This is awesome. I have a perfectly good time capsule that I store movies on and it’s a real pain only having smb1 on it – support being removed on most modern os’s/distribution/devices.
One thing it will now be useful for is a network device for Kodi running on my Amazon FireTV cube. This is neat, because the hardware in the cube seems to decode pretty much everything – including 4K. No need to mess about with transoding malarkey.
I might get another, CEX sell them for 40-50 quid, with a long warranty. Be handy as network backup for my Linux and Windows boxes.
Awesome news!
I loved my airport express, the one that looked like a wall power brick. I used it a lot in hotels when I wanted wired network and you got the airplay bonus for speakers – truly excellent.
Manufacturers should be forced to release the boot loader code of devices when they stop supporting them, so we could keep using the devices after they stop supporting them.
Shiunbird,
It won’t happen, but I agree with this. The argument is always made that it’s unreasonable to expect companies to pay to support products indefinitely, and it’s a valid point. However when a manufacturer stops supporting a product, the community should have a right to support the hardware themselves at our own expense and we shouldn’t have to jailbreak/reverse engineer it. It’s when manufactures won’t provide support and they don’t let the FOSS community pick up that role that we end up with such disappointing outcomes and hardware turning into useless ewaste.
My level of hope for such mandate in the form of a law, at least in Europe, is greater than zero for the very long term. Prices for new hardware keeps climbing, supply chains are constrained, subscriptions galore. At some point, your new bank application will require a new version of your mobile OS of “choice”, and a new phone will be 5000.
Also, heck, I should be able to use some old iPhone touch screen for a personal project, or be able to read the sensors, or whatever that is.
People have been tinkering with cars for ages… reading from the CAM bus, tweaking the injection parameters, adding turbos, changing wheels and tires. It is the same principle (even though I know that car manufacturers are trying to kill it).
Tinkering with my Librem 5 is the first fun I’ve had with a current tech product in years. There’s so much you can do with it. I remember one of the recent interviews with Brian Kernighan and he is saying he carries (I think) an iPhone, a super powerful UNIX-ish computer in his pocket, and he can’t do anything with it, just because he can’t access it.
Shiunbird,
I’m not very optimistic for things improving, if anything we’re seeing things like TVs and cars becoming worse due to enshitification. But if the EU manages to pull it off then I’m all for it.
Yes, there are tons of DIY applications for obsolete hardware. I get pumped about giving new life to what would otherwise become e-waste. Very often though the software create larger obstacles to overcome than hardware integration. This is so unfortunate, yet many manufacturers are guilty of promoting this outcome because non-serviceable hardware translates to higher profits….grr.
Yeah, I’m going to need a new phone eventually and I would like to buy one that fits my values. I’d like to give graphene OS a go, though I do wish more hardware was supported. Most of the phones I’ve purchased were used and it’s a tradition I’d like to continue ideally to do my small part in extending hardware lifespans.
Removing AFP support is actually a good thing. They gave users over 12 years to transition, and over 8 years of use for anyone who bought one of the time machine devices.
Keeping backwards compatibility is DANGEROUS. Even Microsoft dropped SMBv1 support by default in windows 10. There are a huge number of security vulnerabilities relating to backwards compatibility, and many cases in which that backwards compatibility is totally transparent so the user has no idea their connection has been downgraded to a less secure legacy protocol by an mitm attack.
Backwards compatibility is dangerous, and should never be the default. It should either not be present at all, or require jumping through hoops to enable.