Xsan was terrible when I used it. Each side of disks was unique and couldn't be aggregated into one volume, it was slow, expensive, and the software was very unreliable. I can't even count the number of support tickets I opened for this POS.
It did fill a gap for a while unmet by a lot of services that we take for granted now. A Mac Mini with macOS Server to host email, calendar sharing, file sharing, and Time Machine backups went a long way towards meeting a small office's IT needs. It's mostly supplanted by things like Google Workspace, Office 365, Dropbox, and proper MDM solutions these days, but wasn't a bad choice up through maybe 2014 depending on the situation.
My memory of the hosted Outlook/Exchange landscape of that time is much more negative. It was expensive, had inconsistent or costly support for ActiveSync devices, had no integration or federation with existing on-premise Active Directory solutions, management consoles of shared/hosted Exchange providers were difficult to administer. Broadband was much more limited, so remotely-hosted mailboxes were a hassle.
By 2009 at least, iPhones supported hosted ActiveSync. I was writing field service software (“sending people places to do things”) for ruggedized Windows Mobile devices. I do seem to remember some of our customers using ActiveSync from hosted Outlook for emails alongside our software.
I’ve never worked for Apple, but I can only imagine what a disaster their internal IT situation must be given that they use their software nearly exclusively internally but spend almost no effort on enterprise functionality or even basic stability, maintenance and documentation of the limited enterprise features that exist.
The big tech company I do work for uses plenty of Apple hardware, but employs the equivalent of a mid-size startup in teams building internal solutions for gaps in the Apple stack.
You don't need to do that. Buy apps through Apple Business Manager, deploy via MDM. In fact, that's probably a violation of the license agreement, since purchased apps on a single Apple ID are only really supposed to be used by one person (or a family).
Everything in the enterprise env is likely MS based. Active directory, Office, fileshares, etc. I don't even try to integrate Mac's with that stuff, they are on an island.
Not true. I worked for HPE for about 3 years and Apple was one of our largest customers. This was around 2014-2017ish and AFAIK, most of their internal systems were HPE hardware running Linux.
Anyone who has tried to use their server products knows that Apple doesn't use them themselves for running the company. Heck, look at their opensource software and you can tell the teams are using Linux boxes.
Another anecdote: I work for a BigTech cloud provider and our internal applications can’t stand up to nearly the load that our external customer facing services can.
I suspect that every large company's internal IT is a disaster. I say this having worked for a large IT consulting firm that did lots of M&A. Nothing ever got fully integrated, lots of dangling legacy systems.
Well, that comes down to what you think a server is. macOS is definitely unix, and can run server software. You can do that on that rackmount mac. So what makes it not a server?
servers generally will have out of band management, redundant power and networking, some kind of light you can flash to help find it in a rack. stuff like that. oob management is probably the most vital feature of a proper server.
Talking about the hardware here. Yep, Server.app is a shell of what it could be. But I mean the rackmount macs — I was asking GP why they didn't consider those servers ;-)
Not GP, but because they aren’t marketed as servers and don’t have features you’d commonly expect in a server. Video and audio production heavily uses rack mount gear and so having a rack mounted version of a Mac Pro makes perfect sense.
Looking at the back, it is missing things like redundant hot-swapable power supplies. The processor is one oriented toward workstations rather than servers. Not certain if the OS can be configured for huge pages in RAM.
Use case for datacenter Macs is basically just as build servers for Mac and iOS projects, anything else there's not much justification for the hardware cost.
I think it's generally used as a rack-mounted workstation not as a server, guys I've worked for in live video productions definitely seem prefer their gear to be rack-mountable for ease of setup and tear-down, I imagine it's not an uncommon request in other related fields.
Although I can feel my toy vs. necessary alarm going off (you can do live stuff on some truly awful hardware, you don't need the latest Mac even if your personality dictates as much), having stuff rack-mounted is a real help when it comes to keeping track of stuff and keeping your "builds" (literally) reproducible
In a less segmented environment (i.e. non-professional), having gear in a rack can also stop people playing with it.
The rackmount server is really for folks that have their audio / video equipment in a rackmount, not to be a rackmounted server. It would be horrible to use that machine in a server room.
FreeBSD have a lower memory footprint and their network (now together with their filesystem) make it a better server than Linux if you want a better use of your network bandwidth.
I love FreeBSD and would prefer to use it as my desktop instead of Linux, but for servers people should also consider FreeBSD as a great option for a better use of computational resources. (And all this even when Linux have zillions of smart people working hours and hours to optimized it which FreeBSD cannot afford to)
I was also looking into BSDs (FreeBSD and OpenBSD) as an alternative to Linux, but almost every package that I need is available for the top 3 OSs (Win, macOS, Linux) only.
For example, I didn't see vscode for BSD. And I'm worried that maintaining a BSD would be more of a hassle than even Arch Linux.
Back in the day I used run the Linux version of Return to Castle Wolfenstein under FreeBSD (with NVidia drivers), and it ran as fast or faster (per FPS counts).
It used to be that if you were going to run BSD that meant either compiling from ports or source. Packages were only available for the most common components, but the default was ports and most admins went along with that because you got greater flexibility/opportunities to optimize. Back then source compiles of software like Apache were the norm for even the Solaris boxes I worked on. Same for most perl modules (I had a decade-long war with Math::Pari).
Whether it's been added to the binary package build farm or you'll have to build the port yourself I don't know though, but poudriere makes (re-)building ports a pretty pleasant experience.
BSD often does start off feeling like a hassle, but the docs are excellent and once you get a feel for it then it doesn't honestly feel like more work than linux. Note: I run a mixture of FreeBSD and Debian on my personal systems and find them both pretty painless, but I do tend to value "do exactly what I told you to" when it comes to recreational sysadminry so bear that preference in mind when interpreting my thoughts.
> It has been an ongoing sore spot for Microsoft that its highly trafficked Hotmail site runs atop not its own operating system, but the FreeBSD-Apache platform.
> Since it bought Hotmail at the end of 1997, Microsoft repeatedly promised that it would transition Hotmail to Windows NT, then Windows 2000. More than anything, Microsoft's desire was a matter of personal pride. What better way to prove its own contention that NT was just as scalable and robust as Unix than to run its complex, free, Web-based email infrastructure on it. According to the market watchers at Netcraft -- an Internet consultancy based in Bath in the UK -- Microsoft finally has commenced the long-awaited Hotmail migration.
> The name "Hotmail" was chosen out of many possibilities ending in "-mail" as it included the letters HTML, the markup language used to create web pages (to emphasize this, the original type casing was "HoTMaiL").
Those beasts were a weird one. I’d like to get ones and pull its guts out and make it a bit more useful. The roar, the heat, the size and the age make them a little home-unfriendly.
It appears they use at least some amount of Super Micro hardware, possibly in very generic or custom cases (ie, they're not buying Dell.) That being said, I have to wonder how much datacenter space Apple really has. IIRC most of their iCloud services are hosted on various 3rd party cloud providers.
It is hard to say one couldn't see this coming. Of the few features left in Server, the only one I remained interested in was Profile Manager, and that hasn't worked right for a bit now. I'm fine with an Apple that wants to do hardware and operating systems more than software, but I wish they'd make Profile Manager-like features more available for regular people, not just Education or Business users. It would be infinitely useful to have such features for iPhones of family members who aren't so good with tech and could use a heavier hand in making sure the device is up to date and findable when lost. But I digress. I'm glad it's just ending, even if it's barely a surprise.
I worked with a school district who used Profile Manager with ~200 iOS devices. Their in-house tech person got a horrified reaction from an Apple employee at an Apple event when they discussed the district's use of Profile Manager at that scale. The Apple employee seemed surprised that it worked and made it sound like it wasn't really supposed to be used for more than a handful of devices. I thought it was reasonably slick, personally, and would fit the bill for small orgs who otherwise didn't need a subscription MDM service.
Using it several years ago, to me it sure seemed like a reference implementation of an MDM. It just lacked the robustness and flexibility that'd be required pretty quickly once you scale beyond a dozen or two devices.
haha I remember talking to one of the engineers on the enterprise management team at WWDC back in 2015 and he was surprised that anyone used it at all. It really did work well and it did its job.
At least for VPNs, I’ve used tools that autogenerate profiles. In fact, doesn’t Apple have a separate tool to generate profiles IIRC? I guess you’re looking for the remote wipe ability? What would you do for a family member that requires a commercial service?
You could also write your own profile file. It’s really just a policy file written in xml(-ish?) with a poorly documented schema. Or at least it was poorly documented several years ago.
This is very interesting! I need to manage my daughters phones as well as my ex-MiL who I still do tech support for. The Verizon Smart Family stuff barely works and I'm continually having to "fix" it. Can Jamf Now also do Android?
You are a nice ex-SiL for your ex-MiL. I'd love to be able to manage my family's devices in the same way my corporate devices are managed. The built-in OS solutions aren't great (at least on iOS).
I just wish I could dive into my mother's iPad and see the screen.
Even a share screen button inside of FaceTime would be enough. But no... we've got to install some other app and open control center and face east and repeat the ancient incantation of "Holy one, give me the power bestowed upon only Apple Support employees..."
I finally got to try this... it's not very good. Voice quality takes a noticeable dive when screen sharing starts, the frame rate is very poor, and the image gets stuck a lot.
If on iOS 14 or earlier, jailbreaking allows VNC to work. Ofc a non tech person using a jailbroken device likely wouldn’t work so well since they are all only semi-tethered.
Fortunately, if on iOS 15.1 or later, FT allows screen sharing as sibling has also said.
Not surprised, we dropped them the week after apple acquired them and destroyed our processes and controls overnight.
Fleetsmith pre apple was fantastic. I'm concerned that with Apple you're forced into only using App Store apps which simply doesn't work for in-house binaries and or third party tools you don't acquire through the App Store.
Apple likes to do a half in half out dance without consulting with teams that use the tech. Hopefully this doesn't impact too many people.
ABE isn't free, sadly (and iirc you need a DUNS number). Two month trial, but then $2.99 per device. Jamf Now is at least free for the first 3 devices, then $2.00 per month per device after that.
Another idea is perhaps Mosyle Business, which gives you the first 30 devices for free [1].
We were really exited by the Apple Business Essentials plan… until we saw you still needed an MDM. I thought it was an MDM run by Apple.
Looking back and forth between the Profile Manager / MDM capabilities and the new Business Essentials, I have a hard time understanding what the benefit is beyond automatic setup.
Not sure where you saw this. ABE is an MDM in its own right, and can manage device configuration and push apps from VPP. It is otherwise fairly limited, but you don’t need an MDM for ABE.
Not very - there are several reference ones in various languages and frameworks. The difference I'm upset about is that Profile Manager got the closest to acting like a proper enterprise MDM, but still wouldn't let me supervise devices because I'm not a company. I'm not some weirdo who wants full control to devices I don't deserve it on, but I do want to make sure my grandparents' iPhones are up to date, or have the ability to make it beep when they call (from a land line) that they've lost it, or push family calendars or email accounts to their devices so they don't have to enter passwords. Apple has decided those features are only for the enterprise or schools, and I'm sadly neither.
If you're interested this[0] is the MDM reference I'm most interested in these days.
Thanks. That use is very much want I was thinking. I had very similar experiences with elder relatives and it would have made their (our) lives easier and better.
I assumed (perhaps incorrectly) that they meant non-macOS Apple hardware servers.
Also this was never intended for people using an old laptop 'as a server' anyway, of course that works, and people doing that don't need the enterprise features from a server-specific version of the OS.
Well, some people claim that non-Server Windows can't ever be used in a server role, even with third-party software, and that you need a CAL for every person connecting to third-party software on Windows Server, so it's at least possible.
> Except as otherwise permitted by
the terms of this License or otherwise licensed by Apple: (i) only one user may use the Apple Software at
a time, and (ii) you may not make the Apple Software available over a network where it could be run or
used by multiple computers at the same time
It's of course debatable whether, say, using the built-in file sharing counts as "making the Apple Software available", doubly so for third-party servers, but that's exactly the argument some of those Microsoft licencing people are making.
This sounds like they're trying to prevent people running farms of Mac minis AWS style, rather than using them as a server. But a highly paid Apple layer may still disagree.
Apple's current line of ARM chips lack the memory and IO features that are essential for server application. Sure they could be added in a new design but I suspect that the value proposition of in house designs would not as good with the added baggage.
Overall Apple just isn't that interested to produce low margin commodity servers. If Apple had a sliver of intention of returning to the market they would have kept both Xserve and the server OS going with minimal effort. Sony's smartphone business division have been losing money for a number of years but they have repeatedly stated that they will keep it going for the foreseeable future because it would be much harder to enter the market again later without one foot already inside the door.
> Apple's current line of ARM chips lack the memory and IO features that are essential for server application.
Some people just want to run software. There are plenty of cloud service providers that use COTS desktop hardware as servers, and their customers don't complain.
Personally I run a few services on a basic Hetzner Cloud CX11 instance, with memory and IO never having been a bottleneck or concern. If I could wip out a Mac mini to do the same, I'd be happy.
Not everyone needs a Ferrari. Most of us manage quite well with a beaten-up Ford.
I agree. For a lot of people their "server" is just an old laptop running headless in a closet. However eventually one will want more memory and storage, and both are highly constrained on M1 products at the moment.
The demand for high density deployment of Apple hardware has always been present, but the bigwigs from Cupertino could not care any less. On the market one can find all sorts of adapters, both commercial and jury-rigged, for rack mounting a couple of Mac Mini units. And there are companies whose entire business is renting out bare metal Macs in a colo so developers could do their work remotely. I just came across the article below and it made me chuckle - apparently this company went as far as putting dozens of iMac Pro in a rack with the screen facing sideways, and this hotchpotch of gear for rent somehow makes good money.
> If your needs are that mundane, I really don't see why an Apple ARM chip would matter?
I don't think "mundane" is a fair and honest take. Just like most of the industry benefits little to nothing from performance gains from microservices, I'd say that most of the industry benefits little to nothing from high-performance hardware, specially after containerization and container orchestration services took over the industry.
I know for a fact that some WebApps that are household names by now use a single m4.large instance to serve all the traffic from a single deployment region while barely taxing it's CPU quotas. If those quotas were reached it's trivial and boring to just spin up more "mundane" hardware and cease to have a problem.
Why push this myth that everyone needs Ferrari hardware to do everyday work when half the world would do well with today's bicycles?
If we must go with the car analogy then I'd argue that M1 Macs are the Ferrari and Lamborghini of personal computing: They are fast, sleek, and could accomplish a number of specific tasks much better than everything else. Sure it's entirely possible to pull a trailer with a Lambo, however every sane person would agree that any beater truck would have done a better job.
Surprised they didn't recommend Wiki.js under the Wiki section, imo it's the best modern Wiki software right now. Might not be as battle-hardened as Mediawiki but my experience it's more pleasant to use https://js.wiki/
There's a lot to be said for Mediawiki's battle-harened-ness. What makes this so much better, particularly when both could be spun up with just docker containers?
If your identity provider supports LDAP, not really. I was partially responsible for an enterprise wide installation of MediaWiki that talked to LDAP just fine, and this was over ten years ago.
I mean, it has basic support for a private wiki and delegated auth. Just not very robust support. There are a lot of dead ends for common enterprise use cases. It’s just not really designed for this use case, and there other options that do these things better out of the box.
I'm not familiar with this tool in particular but I'm fairly convinced now that wikis are often an anti-pattern.
If you want a wiki as a loosely federated dump of knowledge then a traditional wiki is ok, but if you want it to be instructional material (e.g. "How to build GCC on the Apollo guidance computer") or a cohesive manual, then it should really be generated by something tracked in a proper git repository.
We purchased the Server app back in 2014 because it was an inexpensive and very functional MDM solution for our corporate Apple products. After all, it was only $20 and we could easily run it on a Mac Mini on our network.
It worked incredibly well, but Apple really didn't evolve it much since then and we eventually ditched it entirely in 2017 for Jamf. Since then I've regarded it as a lost enterprise management opportunity for Apple.
Inane? It's a principle of set theory. I'd say that, in response to the other comments, it was fairly apropos for HN to make a set theory reference.
At minimum considering how my comment is common knowledge it is trite rather than inane. So forget about slashdot, let's go old school Usenet flamewar on this, you defend "inane", I'll defend "apropos" & "trite". To be honest I'm not very good flamewars though so maybe you could get things started with a biting personal insult. Something about my genetic lineage or mother's breeding habits is, I think, fairly standard, but I'll a award bonus points for creativity.
Jamf is one of those companies desperately trying to modernize a product that was slow, clunky and poorly designed from the start. Even with their concerted efforts they still put out new software or tools that could have been designed in 2005. Can’t expose your LDAP ports to the internet(!) so their SaaS can connect directly to your domain controller? That’s ok just deploy this ldap proxy software that creates a tunnel for them to route traffic over. And it also needs to have its port exposed to the internet. Apparently having the client establish the connection is out of the question. And that’s not even getting into why LDAP is a requirement for anything but LDAP authentication and AD integrations like AD CA TLS signing.
Some of this might be a little outdated. Haven’t touched anything like that in ages.
Does that really matter? There are really good ARM processors right now, and biggest advantages of M1 - power efficiency - really matter more on the desktop than on server.
It's exactly the opposite. A reduction in power can save millions of dollars in energy and cooling. All of the constraints and challenges of building a desktop are scaled up. This doesn't even factor in the savings in space. Real estate isn't cheap.
A small, power efficient and quiet home server is something I’d love - extra points if the power supply is built in.
I’ve tried a lot of Intel Nucs and they come close. The Nuc9 especially so, as it fits PCI cards.
It’s close and the 10gb networking makes it attractive.
However it maxes out at 16gb ram (a Nuc 9 goes to 64) and virtualisation isn’t quite there yet I don’t think. Fusion says it supports the M1 but I don’t get good results.
I’d like esxi or Proxmox on it, but could learn to live with it as is I suppose.
Runs Debian (Armbian). Provides file shares over SMB, Syncthing, S3 (Minio). Also does some monitoring of other systems via Prometheus scraping of node-exporter. Everything except Samba runs in containers.
With two mirrored 6TB HDDs it has no problem keeping up with the household 1GbE network. Quad A55 cores, 4GB with Zram and it has all the power I need.
After messing with some RPis and Odroids, I went the other way and just got a single used Lenovo workstation off Ebay. After a memory upgrade (with ECC) and adding an SSD for the system disk, it was about $200, or about the same as fully outfitting two higher-end RPi4s. Easily runs a half-dozen dockerized services, with capacity to spare, and takes care of a bunch of ZFS-formatted, mirrored, spinning rust internal hard drives (ZFS on USB disks is... not fun).
It can't live-transcode video above about 720p without stuttering, but then, most of those ARM boards can't either, so I just have to make sure whatever client I'm using can handle native formats for all my media.
Not sure how it is on power use, but I much prefer managing all that stuff on one machine.
Typically I like to use my old laptops as "servers".
They usually get dropped or ran over or screen hinges broken or something like that. Long before the actual hardware inside gets a chance to fizzle out.
They have a lot of advantages in that they are low power, have built-in UPS, small, have a keyboard and display. The disadvantage is that you don't get a lot of storage options. I have some USB 3 attached storage, but I wanted something a little different this time.
Most recently my RPI 4 8GB serving as media server got replaced by my girlfriend's Asus laptop. She knocked it off a table and the corner of the display got smashed in. Now it is running Linux and Kodi, among other things. Can barely see it sitting there under the TV.
A bit better at 4K out then RPI 4, but not by a whole lot. Don't do any transcoding on it, though.
My "Big Servers" for my home lab are 3 ancient HP 1U rack servers. Built a small horizontal rack so they "hang down" instead of sideways and they take up almost no room. They run some high efficiency Xeon processors so I can get away with running them off of one household outlet. I only leave one on and spin up the other two when I need the extra capacity for some project.
Dual socket 8 cores and 72GB of RAM each.
Figured out that if you go and look at Vmware hardware support and get a server that is _just_ one generation older then they are willing to support than these things are dirt cheap on the used market.
If they can't run the latest version of Vmware then nobody wants them. Linux don't care, of course.
I think I spent a total of 600 dollars for all 3 servers, but it could of been a lot less. It's been a while.
I hate dealing with server-grade hardware because the lights out stuff is insecure (100% get one with a dedicated LO network port), dealing with firmware configuration is tedious, take forever to boot up, and dealing with the raid controllers is misery. But it's the cheapest way to get a ton of capacity. Especially if you have a project that wants to use IPMI and such things.
It matters for targeting "Apple Silicon" (aka M1, for now) Macs. It's complex to build a binary that works at feature parity on M1.
Licensing requirements for VMs and lack of server hardware that can be put in datacenters are both obstacles to target Apple Silicon. Not insurmountable, you can build enclosures for a rack of Mac Studios & Mac Pros and orchestrate VMs with Hypervisor framework, but it's a large enough obstacle that major cloud providers don't offer it and that pushes large costs onto the CI providers, which then get pushed to the folks trying to build tooling that works on M1 Macs. That includes the open source ecosystem.
The alternative is cross compilation, either cross-arch or cross-platform or both. All of these options are brittle and complex.
Speaking from experience here, I've had to follow issue trackers for: programming language teams shipping parity for M1, major libraries (think of foundational libraries like GRPC's libraries for each language) not yet shipping binaries for M1, and developer tools and OSS teams producing binaries. All of this needs to come together, and it hasn't yet.
The whole ecosystem is straining under the weight of this complexity to target engineers' laptops.
It's been nearly two years since the Apple Silicon kits first shipped, and the only instances you can purchase from AWS are metal instances, at approx. $10k/year/instance.
That's not accessible to open source/free software projects and not affordable or efficient to utilize for small projects and highly intermittent workloads like CI.
And the minimum lease time is 24 hours, making it inaccessible to hobbyists or on-demand, intermittent work.
I wish MacOS licensing were available for Graviton and ARMv8 generic, that would be incredible. It isn't, and the costs to support Apple's platform is a substantial barrier for many projects.
Ridiculous. You can buy an M1 Mac mini for $699 (MSRP) and often for $599 from resellers. To develop a Mac or iOS app, you need a Mac, so you're not spending more.
Hobbyists? I spend $700/year to have access to a maker space. I spend that much on woodworking tools annually (if not more). Hobbyist doesn't mean free.
Operations/watt, which is a function of power efficiency (i.e. how many floating point tera-ops can you do in 1 second), is the primary driving performance criteria and engineering metric for all modern desktop, server, and mobile platforms. There is almost no metric which matters more when it comes to high performance compute.
That said, Apple isn't going to sell chips these to anyone else, period. And everyone right now is hustling to match the same levels of efficiency in newer designs, and they'll get there sooner or later. So "Apple Servers" are a completely moot point, but power efficiency? Yes, it literally matters more than ever, across every spectrum of the industry, and it's why the M1 was something people cared about anyway.
When the words Apple and Cloud come together I start to twitch.
It’s done to death that their web services are of dubious quality, and they have got better. However you don’t have to go very deep or use services for very long to find some very sharp edges.
Maybe I'm just seeing what I want to see, but I get the impression that Apple's approach to developing products can lead to very good highs but pretty poor lows considering the amount of money they could easily invest, I would also be somewhat hesitant with an Apple Cloud unless they had something actually new to bring to the market other than basically rent-seeking for people who want to develop (say) CI pipelines for their products.
It probably won't happen, but if Apple did want to get into the server market, it would make a lot more sense for them to support linux on their hardware for server use cases than to make their own server software.
When apple sold Xserve the support was excellent. Once I called and they had the guy who wrote the code go into a server room and duplicate an unfinished problem while I was on the phone.
If they did go into this business seriously they’d probably do this again. That kind of support doesn’t scale for consumer products.
Yeah, I know of a company that was using a first gen xserve up until last year. Something broke and they couldn't get (or couldn't justify the price of) replacement parts, but nearly two decades of use is very impressive.
> this is the definitive signal that will not happen.
…that it will not happen via an add-on “server” program.
But TBH it’s hard to see how servers could be much of a market for Apple. People use the cloud more than local (“departmental”) servers these days and the cloud back end belongs to Linux, typically with a provider on top (AWS etc). Apple’s strengths lie elsewhere.
What Apple need to do now is make their devices more linux-server friendly out of the box.
I have scripts set up on all of my storage servers to scan for and remove .DS_Store files, as it essentially doubles the amount of files on a server which for indexing is incredibly bad. My complaints to the Mac users were usually met with replies such as 'Well if you ran Mac server that wouldnt be a problem'. Now they dont have that arguement to stand on.
I am aware of the command to stop apple devices writing these files to network shares, but regularly our mac users 'forget' to run this command after setting up new profiles or upgrading OSs etc.
You hit the nail on the head. It provides a useful service to 'Mac' users. To literally every other OS user these files are considered bloat and slow the system down for everybody. Hell, linux users performing searches on these directories get duplicate results for every file a mac has touched.
I appreciate it may come across as a bit hostile towards the users, but in reality in a budget constrained environment where we cant make it perfect for everybody, we must make sacrifices to make the majority of users have a better experience. In my opinion mac users not having metadata of when a file was last modified pales in comparison to the rest of the business searches taking twice as long on samba shares.
Thankyou for correcting me. I was speaking from old knowledge as this was a while ago I set this up and couldnt quite remember exactly how the .DS_store system worked. I thought it was one .ds_store file per file.
That being said, depending on your business and data structure this doesnt detract from the point.
Also if you have a deep nested directory structure with no real files in it, would every level contain a .ds_store file?
The .DS_Store file contains macOS Finder's view preferences for the directory. Finder will only write a .DS_Store file to directories where a user has actively altered their view preferences for that directory (e.g., switched from icon to list view, “cleaned up”/rearranged icons, etc.). Just navigating through the directory doesn't create a .DS_Store file.
There is one .DS_Store file per directory. It contains information like window size, icon position, folder background, thumbnails, etc. Deleting/vetoing .DS_Store files will not hurt anything of substance - other than discarding user preferences.
There are also AppleDouble (._) files, which are one-per-file. These contain the file's extended metadata and resource fork. Deleting these _may_ cause data loss if there's anything important stored in the resource fork. A better option is to enable vfs_streams on your Samba server to allow storing the additional forks natively on your filesystem (e.g. as xattrs).
(If you're using a modern Windows file server, I believe the resource fork is automatically mapped to an NTFS alternate data stream.)
There is also the (evidently dying) principle of by default wanting control over my computer and filesystem. I know this is not compelling to most people anymore but I feel like if I want 25 files in a directory, I expect to see 25 there. Not 26 because the operating system really really really really wants to pollute it with one more file. I want 25 there. If I wanted that other file there, I would have explicitly commanded my computer to put it there.
I also object to my operating system running all these background processes on MY computer without me commanding it to, and it suggesting on its own that I do this or run that, again in absence of any command to do so. More and more, operating systems and applications are treating MY computer as a dumping ground and science experiment: for things it wants to do, instead of what I want it to do.
I should not have to go off and find a setting somewhere just to stop my operating system from doing things on its own I don’t ask for.
It's an impedance mismatch. You're moving files between a system that supports multiple streams per file (data fork + resource fork) to a system that has no concept of different streams (POSIX). The extra stream has to go somewhere, or you get data loss.
It's worth noting that almost all modern file systems support multiple streams. NTFS has alternate data streams, Ext4 has xattrs. Modern SMB and NFSv4 also both support this at the protocol level.
The problem arises when you're using Samba (without vfs_streams enabled), or you're writing to a legacy FAT filesystem, in which case you start getting the AppleDouble files - again, to prevent data loss.
For plenty of users "control over my filesystem" means being able to put an icon in a certain position and have it stay there when they come back to the folder.
I believe you're thinking of the AppleDouble files with name ._file. Those haven't really been used for the past 10 years or so (unless you run software older than that, obviously).
> Also, these files provide a useful service to Mac users.
With respect to DS_Store files on shared network drives, that is not true. Such files provide utility to a single Mac user, whichever uploaded the DS_Store file to that directory last. This file is used to store user preferences, which breaks as soon as there are two or more users with different preferences. Simply put, it does not belong on shared network drives at all.
> Simply put, it does not belong on shared network drives at all
My home NAS file shares are used by only me, so I'd like it to be supported there. So at the very least it needs to be configurable per-share. And in that case, having the server admin add them to veto lists seems like the easiest solution.
Sure, but not on shared network drives. A network drive only used by a single person is another matter. (And probably not very common, though I also have one.)
Assuming you’re running a smb server, you could just veto the files. Windows isn’t much better since it likes to create thumbs.db almost everywhere too (which I also veto, but vetoing them can increase the load and bandwidth requirements and your server and clients)
Yeah Vetoing is an option, although without testing I do not know how the mac clients would react on saving a file and its metadata was not allowed. Would the mac throw an error?
EDIT: I have had people say this to me before about windows and thumbs.db. But I personally have not seen this in the wild. Maybe its what old versions of windows did and people are still remembering this?
I haven’t seen any errors and macOS seems to handle it greacefully. You can also disable it on macOS clients for network servers individually but that seems to be a loosing battle (even if you control all clients). They are finder settings after all
Thumbs.db files are created on my windows 11 pc at least. They’re only created for files that have metadata that requires reading those files. Explorer likes to display the metadata (sometimes) for some folders that have a lot of media in it (pictures, music, videos, etc). If the thumbs.db file is missing, windows will partially read every media file on the server to show thumbnails, that obviously creates unnecessary load but it’s really a trade off that might not make sense for most.
LOL - for fifteen years Apple has refused to allow an ext4 formatted volume to auto-mount on a MacOS X machine.. how hard is that? Apple is non-Apple hostile, and proud of it
As far as I know, MacOS doesn’t support ext4 at all (I vaguely recall some read-only ext4 FUSE thing, but obviously that doesn’t help if you need to write).
There's also fuse-ext2 via macFUSE but its doc still say: "Even though write support is available, please do not mount your filesystems with write support unless you have nothing to lose."
It also doesn't seem surprising to me that Apple itself wouldn't support ext2/3/4. It's something very few Mac users are going to care about. Why would Apple dedicate any resources to it?
FWIW, I've been using Linux and Macs for over two decades and I can only recall maybe a single time I even wanted to mount an ext formatted drive onto a Mac, and vice-versa (HFS onto Linux). For exchanging files via floppy disks or thumb drives, various flavors of FAT have been near universal for ages. CDs/DVDs/BluRay use their own filesystem. And FTP/NFS/SMB/AFP/HTTP/SCP/RCP have been around forever for exchanging over a network. (Before ssh/scp, there was unencrypted rlogin/rsh/rcp... I'm dating myself.)
The problem is, there is still no somewhat decent way to have a USB drive that's readable and writable by all three major OS families (Windows, Linux and macOS) and keep UNIX permissions working on it:
- FAT32 can't go above 4GB files, can't do sparse files and can't store UNIX permissions at all
- NTFS can do large and sparse files, but can't deal with UNIX permissions, instead bringing NT ACLs (which no one else understands). exFAT used to be patent encumbered and has the same limitations that NTFS has, plus its support is nowhere near as battle-tested as NTFS is
- HFS/APFS are not supported at all under Windows and support in Linux is spotty at best
- ext4 has third-party read (and if you're risk-tolerant, write) support on Windows and Linux, but it's hacky or expensive
> Why? Who needs this? I don't even understand the use case for file permissions on an external drive.
Shoving data across operating systems using USB sticks or drives is a pretty standard thing to do, at least in the corporate world which is filled with slow internet connections (in related words, never underestimate the bandwidth of a truck filled with SD cards), and particularly when having to send data to clients or vendors.
The fact that there is no standard that works on all three major OS families and in the best case also supports encryption is maddening - macOS has Filevault, Windows has Bitlocker and Linux has LUKS, and neither of the three can understand any other.
Just ask an IT helpdesk in any major company how many questions they get a day by people who have to transfer files to a vendor or client, with encryption. Extra fun in anything involving creative work because most of that happens on Macs whereas most big-corp customers use Windows because of Active Directory and GPO. (And no, ZIP isn't an answer because Apple's built-in unarchiver routinely gobbles up when encountering password protected ZIPs)
> I don't even understand the use case for file permissions on an external drive.
Unlike Windows which conveys the fact that an executable is an executable by the file suffix .exe (or .cmd/.bat, and I think *.js was at least until Windows 7 associated with the JScript interpreter), Linux and macOS do so by using the execute permission on the file.
The use case would be to copy a Java application, let's use jadx, downloaded and extracted on a Mac, then copy it via an USB stick or a Windows CIFS file share on another Mac or a Linux machine, and have the command bin/jadx-gui still start up the UI when clicked upon. As soon as any non-Unix filesystem is in the copy path, it gets inevitably nasty.
I mean, it doesn't solve the filesystem bit but instead of zip, what about 7zip?
There's a version of 7zip for windows, linux and Mac. It's supposedly safe for now. Until the next vulnerability is found I guess but that is the same for every option.
Bit of a downside that it is a third party option that would then have to be installed everywhere you want it rather than a native option though.
>never underestimate the bandwidth of a truck filled with SD cards
Latency is the issue! Once I was working at a museum and we needed to get these stupidly high detail 3D scans to the university. Considering the IT policy and available bandwidth it was quicker for us to put all the files on a single HDD and walk it over.
Do you find yourself completely baffled when you're advocating for something as simple and mundane as interoperability with desktop computers and USB drives in 2022 and people are arguing against it?
I'm not arguing against it. I'm trying to understand the use case. There is already interop using FAT/exFAT. Apparently the problem with exFAT is not having an executable bit in order to run a Java binary.
File permissions usually don't make sense on an external drive that's moving between computers, much less computers running a different OS.
I just don't see why it's surprising that Microsoft and Apple don't have built-in support for ext when FAT/exFAT support is good enough for 99% of interop use cases.
FAT has serious limitations wrt filesizes, it is not good enough for interop use. Even Microsoft's own Windows install media has a problem with it.
exFAT is better, but it is also quite recent development on Linux, since it was licensed and only in 2019 that Microsoft publicly said, they would not ask for any licensing and the specification became publicly available.
Both of them are quite unfriendly to flash memory, but exFAT is good enough.
What about UDF? That's what I use on my USB drives that are going to be shared amoungst all 3 OS', and support is excellent with many decades of testing.
Files up to 16 EiB, supported from Windows XP and up, Mac's with OSX and up, Linux had some bugs back in 2010 which prevented you from dealing with very large devices that were over 80% full.
Been doing it for over 10 years now, and it is still one of the best formats for universal access.
Dont forget that exFAT is not journaled which is the biggest issue by far.
Really it would be nice if everyone could just natively read EXT3/4 out of the box - theres no reason NOT to add it. And even though Microsofts new file system still doesn't seem to have taken off (ReFS) it is quite good.
> Why does Windows userbase care more about Linux drive interop than Mac userbase does?
Mostly it doesn't. And in any case, Windows doesn't ship with native support for mounting ext either. Only recently by installing WSL can it do so. So a very tiny minority of Windows users care about it.
How hard is it to maintain and keep secure an entire modern, complex, evolving file system which the OS itself does not even use (so nothing like simple, frozen, and interchange-standard FAT) in an operating system kernel?
Very hard.
Instead there are 3rd party implementations. Those are independently maintained, serving directly the (comparatively small) target audience.
What I'd love is a robust, modern, relatively easy-to-configure, focused DAV server— specifically one that includes calendaring and notes. Everything out there is either incomplete, fussy as all getout to set up, dogmatically simple and therefore requires external mitigation, only supports one or two clients, and/or some janky PHP amalgam.
What I'd REALLY love is to be able to run all of my own services in place of iCloud. I realize that this stuff isn't free to develop and I'd happily pay for it. They've been moving further away from standard formats and protocols with no way for others to integrate in the same way and I'm curious if there are any legal anticompetitive actions on the table should they continue to do so.
Before anybody says it— I have business needs that favor using MacOS directly on Apple Hardware and as a result, iOS is a good choice. Moving from Apple would regularly cause me more grief than these things do. No, you don't know my use case better than me and I have exhaustively explored all alternatives both ancient and modern.
I've found that Radicale works very well for DAV calendar, tasks, address book. I combine it with Syncthing for files, and use Obsidian.md as the note-taking application (using the synced folder for storing the notes folder). Of course when you have a synced folder you can use whatever note-taking application/system you want.
I'm not sure if that will work on iOS since iOS doesn't really have a filesystem - there are apps available on iOS for Syncthing and Obsidian, but I don't know if Obsidian will be able to access the Syncthing synced folder. But it works great for me on Linux PCs + Android. Much better than Nextcloud which is too complex and is a pain to administer over time. Syncthing is super simple to set up does file sync very well, much better than Nextcloud does. Radicale is also easy to set up and just works.
iOS only needs to be a client and it has standard CalDAV/CardDAV clients built into the OS. I've used Radicale— it works OK for very basic functionality, but doing something as simple as making a shared calendar involves creating one in the GUI, logging into the server and sym linking them between user directories.
The nice thing about the groupware functionality in Apple Server is that it used all of these standard protocols so it was entirely interoperable with other devices AND it had a nice smooth administration experience.
At the moment, I just pay $15/mo for Cloudron which handles email, is decently smooth for administration though a little more disjointed between the apps than I'd prefer, and can "one click" deploy Radicale, NextCloud, Sogo, et. al. I used to administer servers but it's not what I do now and I have no interested in sinking non-work time into work-like tasks.
I believe you're confusing the .DS_Store files which are per-directory and used for storing Finder window metadata with the "._" prefixed files (so-called AppleDouble format) which are used to store a file's resource fork. (These also show up in zip files created via the Finder.)
You can configure Samba to store resource forks using xattrs instead. See fruit:resource:
You can veto the .DS_Store files. The consequence to Mac users is that the Finder won't remember any display changes they make to windows that correspond to network folders.
> You can veto the .DS_Store files. The consequence to Mac users is that the Finder won't remember any display changes they make to windows that correspond to network folders.
Good. Why would one user's view preferences have an effect on another user? That is what happens if .DS_Store files are left on the server.
It makes a weird sort of sense if you consider the folder to be like a stockroom - when you tell someone the box they want is “on the right as you enter” you expect it to be in the same place you saw it.
When I used a Mac at work .DS_Store files on shared drives were annoying. It meant directories loaded with someone else's preferences about how it should be shown.
Apple apparently has forgotten that it makes Finder, but that should have changed to a .sqlite file in the user’s preferences approximately fifteen years ago. It’s so annoying.
> At the end of the month it sums up to a complete employee pay check in addition to the employee pay check.
Sure, there are some large companies that the total cost of subscription services would add up to the equivalent of 1 more employees paycheck but is that really so crazy? For what they get out of it in terms of not having to pay multiple IT people it seems like a pretty decent deal. Someone has to maintain/update/support/etc the tools, it's not free.
Mobility and security concerns changed the game. “Zero trust” / BeyondCorp is a better architecture but requires competencies that are very hard to hire for. It makes it easier to justify the cost of outsourcing the backend to Microsoft or Google.
Interesting. I suppose the strategy is to get everyone to use Apple Business Essentials (https://www.apple.com/business/essentials/), which isn't yet available outside of the US.
It's disappointing they're discontinuing MDM services as well. I've been using Profile Manager and Apple Configurator to manage personal devices + family devices across my extended family. Looks like I should investigate Jamf Now as a replacement based on this thread.
Three things that happens to be what I thought important apart from MDM, gets no mentions and I assume no use at all? Caching Server used to cache iCloud content as well. Never tried it with iCloud Photos though so I am not sure if it is a backup solution if Apple ever log you out of APPLE ID.
This basically ends the dreams of Apple ever selling a Time Capsule with iOS backup and NAS functions.
> The most popular server features—Caching Server, File Sharing Server, and Time Machine Server are bundled with every installation of macOS High Sierra and later, so that even more customers have access to these essential services at no extra cost.
It's too bad that, assuming that documentation is still up to date, Apple doesn't allow smaller IMAP mail providers to integrate with the push notification service.
You can in fact get a certificate for 99 per year and use that to generate entitlements that Apple will happily sign CSRs for. It's how we use MicroMDM.
We would love to be able to send push notification to native apple Mail app for our users but we never wrapped our head around the whole certificate thing. (And Apple was... unhelpful).
Would love to have support on this! Are you available?
Let me know (nicolo.benigni@qboxmail.com)
I think the best explanation (including APNS which is on top of IMAP open connections) on the certificates and key usages is described here: https://micromdm.io/blog/certificates/
ABE and ABM both use the same technology so it's pretty much exactly how Apple intends it to be used. There used to be someone hosting a shared CA for this, but it because cumbersome to deal with third parties not loading the CRL and not using OSCP.
It happened to us with Fleetsmith too, it was great then overnight everything stopped working, which incidentally was because apple acquired them. I like Jamf'd better though, it's going into the lexicon!
Either we were coworkers, or every company that uses JAMF ends up with that saying. In fact, we made a slack response that would post a gif of Jeremy Jamm from Parks and Recs saying "You just got JAMFed" whenever anyone mentioned JAMF.
> I worked at a company that used the phrase "getting jamf'd" as a verb to describe when the management system broke things on your macbook.
It's just a management framework - Do developers who use windows say "getting sccm'd" or "getting Intuned" when SCCM/Intune does management actions on their devices?
> I got Jamf'd pretty hard this morning and now my build is broken...
If the IT department is pushing stuff out through the management framework on the devices that is "breaking builds" - they are doing device management wrong.
And if the IT department is pushing out Config Profiles which break things people use every day, thats more a people problem than technology/Jamf problem. (as in the IT dept are morons)
I think part of the problem is that many IT departments have a culture which focuses on the needs of less-technical users. They often don't have a good appreciation of the more demanding needs of technical teams. Non-technical users need a lot of handholding – an approach which can be irritating to the technically advanced. A locked-down UI which doesn't let you change any settings may be the right approach for non-technical users, developers can find it infuriating that you won't let them try to solve their problem themselves, and instead force them to talk to a helpdesk who don't understand it either and want to follow some script ("Have you tried rebooting?"), before they let you talk to someone who actually understands what is going on
Maybe software like Jamf isn't inherently a problem, but it can encourage IT departments with that kind of culture to do more things to irritate developers (like automatically run buggy scripts on developer laptops without any notice or easy visibility into what those scripts are and what they do), which without that kind of management software available to them, they would have been less likely to try to do. Developers are more likely to have configured their environment in custom ways which will cause IT's buggy script to play up and cause problems
I used to work for Oracle, and while I was there Oracle had a whole separate IT department just for the engineering/R&D org, while the main IT department serviced the rest of the business (sales/support/finance/legal/HR/etc). I don't know whether Oracle still uses that setup, but (if the company is large enough to sustain it) that could be an approach to avoid some of this
> It's just a management framework - Do developers who use windows say "getting sccm'd" or "getting Intuned" when SCCM/Intune does management actions on their devices?
Ubisoft Massive does have the phrase "got GPO'd" if that helps.
A lot of the features in macOS Server are now built into the default install for MacOS (caching server etc.) while others were discontinued long before the latest version of server came out (Wiki Server etc.)
Yeah but they say Time Machine Server is an option in sharing, but it has never appeared in any of my computers at any point in time. Apple's gonna Apple.
376 comments
[ 2.4 ms ] story [ 321 ms ] threadThe big tech company I do work for uses plenty of Apple hardware, but employs the equivalent of a mid-size startup in teams building internal solutions for gaps in the Apple stack.
This isn't true.
If MacOS Server is being discontinued, are those being orphaned? Or can you run the regular OS headless?
[1] https://www.apple.com/shop/buy-mac/mac-pro
Native development shops running in-house build pipelines I imagine. But even that has completely gone to the cloud in the last 5 years.
https://www.apple.com/shop/buy-mac/mac-pro/rack
https://www.intel.com/content/www/us/en/products/details/pro...
10.6 (Snow Leopard) was the last version where they sold a separate more expensive server edition of the OS.
As for server hardware, I have to imagine the 3rd party 1U Mac Mini rack enclosures are already more popular than Apple's official rack mount Pro. https://www.sonnettech.com/product/rackmacmini.html
Use case for datacenter Macs is basically just as build servers for Mac and iOS projects, anything else there's not much justification for the hardware cost.
In a less segmented environment (i.e. non-professional), having gear in a rack can also stop people playing with it.
https://web.archive.org/web/20021021164226/http://www.securi...
I love FreeBSD and would prefer to use it as my desktop instead of Linux, but for servers people should also consider FreeBSD as a great option for a better use of computational resources. (And all this even when Linux have zillions of smart people working hours and hours to optimized it which FreeBSD cannot afford to)
For example, I didn't see vscode for BSD. And I'm worried that maintaining a BSD would be more of a hassle than even Arch Linux.
* https://wiki.freebsd.org/Linuxulator
Back in the day I used run the Linux version of Return to Castle Wolfenstein under FreeBSD (with NVidia drivers), and it ran as fast or faster (per FPS counts).
* https://en.wikipedia.org/wiki/Return_to_Castle_Wolfenstein
https://www.nvidia.com/en-us/drivers/unix/freebsd-x64-archiv...
someone at Nvidia must be a FreeBSD stalwart and I’m okay with that
Whether it's been added to the binary package build farm or you'll have to build the port yourself I don't know though, but poudriere makes (re-)building ports a pretty pleasant experience.
BSD often does start off feeling like a hassle, but the docs are excellent and once you get a feel for it then it doesn't honestly feel like more work than linux. Note: I run a mixture of FreeBSD and Debian on my personal systems and find them both pretty painless, but I do tend to value "do exactly what I told you to" when it comes to recreational sysadminry so bear that preference in mind when interpreting my thoughts.
> It has been an ongoing sore spot for Microsoft that its highly trafficked Hotmail site runs atop not its own operating system, but the FreeBSD-Apache platform.
> Since it bought Hotmail at the end of 1997, Microsoft repeatedly promised that it would transition Hotmail to Windows NT, then Windows 2000. More than anything, Microsoft's desire was a matter of personal pride. What better way to prove its own contention that NT was just as scalable and robust as Unix than to run its complex, free, Web-based email infrastructure on it. According to the market watchers at Netcraft -- an Internet consultancy based in Bath in the UK -- Microsoft finally has commenced the long-awaited Hotmail migration.
> The name "Hotmail" was chosen out of many possibilities ending in "-mail" as it included the letters HTML, the markup language used to create web pages (to emphasize this, the original type casing was "HoTMaiL").
* https://en.wikipedia.org/wiki/Outlook.com#Launch_of_Hotmail
Those beasts were a weird one. I’d like to get ones and pull its guts out and make it a bit more useful. The roar, the heat, the size and the age make them a little home-unfriendly.
https://www.macrumors.com/2017/02/23/apple-ends-relationship...
https://9to5mac.com/2021/02/12/super-micro-spy-chip-story/
The rest of it I would imagine is just commodity hardware like you said (supermicro is good stuff generally)
But it is also known they moved a lot of storage to Google servers.
Rhapsody came and gone.
https://en.wikipedia.org/wiki/Copland_(operating_system)
It was the code name for early macOS, back when it was mostly a Mac OS-skinned OPENSTEP.
We unwisely used it in production to host NetInfo and various other services.
Rhapsody isn’t dead ;-)
Avie is still blessing us all.
Shame that they killed Profile Manager though. Not that it matters too much.
Even a share screen button inside of FaceTime would be enough. But no... we've got to install some other app and open control center and face east and repeat the ancient incantation of "Holy one, give me the power bestowed upon only Apple Support employees..."
Thanks!
Better than nothing, but still pretty bad.
Fortunately, if on iOS 15.1 or later, FT allows screen sharing as sibling has also said.
Check out Splashtop SOS for remote viewing iOS and iPad OS devices:
https://www.splashtop.com/remote-access-view-iphone-ipad-ios...
https://apps.apple.com/us/app/splashtop-sos/id1230853703
https://www.apple.com/business/essentials/
Fleetsmith pre apple was fantastic. I'm concerned that with Apple you're forced into only using App Store apps which simply doesn't work for in-house binaries and or third party tools you don't acquire through the App Store.
Apple likes to do a half in half out dance without consulting with teams that use the tech. Hopefully this doesn't impact too many people.
Another idea is perhaps Mosyle Business, which gives you the first 30 devices for free [1].
[1] https://business.mosyle.com/pricing
Looking back and forth between the Profile Manager / MDM capabilities and the new Business Essentials, I have a hard time understanding what the benefit is beyond automatic setup.
Does anyone else understand the business case?
Honest question. Im a biologist not at all a software guy but have ‘managed’ low level IT at startups so I think I get the value prop.
If you're interested this[0] is the MDM reference I'm most interested in these days.
0 - https://micromdm.io
Also this was never intended for people using an old laptop 'as a server' anyway, of course that works, and people doing that don't need the enterprise features from a server-specific version of the OS.
And, indeed, Apple does have a similar clause in their EULA: https://www.apple.com/legal/sla/docs/macOSMonterey.pdf
> Except as otherwise permitted by the terms of this License or otherwise licensed by Apple: (i) only one user may use the Apple Software at a time, and (ii) you may not make the Apple Software available over a network where it could be run or used by multiple computers at the same time
It's of course debatable whether, say, using the built-in file sharing counts as "making the Apple Software available", doubly so for third-party servers, but that's exactly the argument some of those Microsoft licencing people are making.
Overall Apple just isn't that interested to produce low margin commodity servers. If Apple had a sliver of intention of returning to the market they would have kept both Xserve and the server OS going with minimal effort. Sony's smartphone business division have been losing money for a number of years but they have repeatedly stated that they will keep it going for the foreseeable future because it would be much harder to enter the market again later without one foot already inside the door.
Some people just want to run software. There are plenty of cloud service providers that use COTS desktop hardware as servers, and their customers don't complain.
Personally I run a few services on a basic Hetzner Cloud CX11 instance, with memory and IO never having been a bottleneck or concern. If I could wip out a Mac mini to do the same, I'd be happy.
Not everyone needs a Ferrari. Most of us manage quite well with a beaten-up Ford.
The demand for high density deployment of Apple hardware has always been present, but the bigwigs from Cupertino could not care any less. On the market one can find all sorts of adapters, both commercial and jury-rigged, for rack mounting a couple of Mac Mini units. And there are companies whose entire business is renting out bare metal Macs in a colo so developers could do their work remotely. I just came across the article below and it made me chuckle - apparently this company went as far as putting dozens of iMac Pro in a rack with the screen facing sideways, and this hotchpotch of gear for rent somehow makes good money.
https://www.datacenterdynamics.com/en/analysis/not-just-stac...
I don't think "mundane" is a fair and honest take. Just like most of the industry benefits little to nothing from performance gains from microservices, I'd say that most of the industry benefits little to nothing from high-performance hardware, specially after containerization and container orchestration services took over the industry.
I know for a fact that some WebApps that are household names by now use a single m4.large instance to serve all the traffic from a single deployment region while barely taxing it's CPU quotas. If those quotas were reached it's trivial and boring to just spin up more "mundane" hardware and cease to have a problem.
Why push this myth that everyone needs Ferrari hardware to do everyday work when half the world would do well with today's bicycles?
https://www.carscoops.com/2015/05/who-said-you-cant-use-lamb...
If you want a wiki as a loosely federated dump of knowledge then a traditional wiki is ok, but if you want it to be instructional material (e.g. "How to build GCC on the Apollo guidance computer") or a cohesive manual, then it should really be generated by something tracked in a proper git repository.
It worked incredibly well, but Apple really didn't evolve it much since then and we eventually ditched it entirely in 2017 for Jamf. Since then I've regarded it as a lost enterprise management opportunity for Apple.
At minimum considering how my comment is common knowledge it is trite rather than inane. So forget about slashdot, let's go old school Usenet flamewar on this, you defend "inane", I'll defend "apropos" & "trite". To be honest I'm not very good flamewars though so maybe you could get things started with a biting personal insult. Something about my genetic lineage or mother's breeding habits is, I think, fairly standard, but I'll a award bonus points for creativity.
Some of this might be a little outdated. Haven’t touched anything like that in ages.
You can have a 5U server with lower power usage that is a LOT cheaper than 1U power hog.
Obviously at AWS scale this is abstracted away, but I think you are still paying a lot for the power you consume.
So not being able to specifically use Apple silicon isn't a big deal.
However it maxes out at 16gb ram (a Nuc 9 goes to 64) and virtualisation isn’t quite there yet I don’t think. Fusion says it supports the M1 but I don’t get good results.
I’d like esxi or Proxmox on it, but could learn to live with it as is I suppose.
Running a GUI when headless seems a bit crazy.
There's many hundreds of thousands of windows servers that run a GUI.
Not saying I like it; but, it's a thing and nobody seems to care that much.
This is what I use for my file server:
https://ameridroid.com/products/odroid-hc4
Runs Debian (Armbian). Provides file shares over SMB, Syncthing, S3 (Minio). Also does some monitoring of other systems via Prometheus scraping of node-exporter. Everything except Samba runs in containers.
With two mirrored 6TB HDDs it has no problem keeping up with the household 1GbE network. Quad A55 cores, 4GB with Zram and it has all the power I need.
It can't live-transcode video above about 720p without stuttering, but then, most of those ARM boards can't either, so I just have to make sure whatever client I'm using can handle native formats for all my media.
Not sure how it is on power use, but I much prefer managing all that stuff on one machine.
They usually get dropped or ran over or screen hinges broken or something like that. Long before the actual hardware inside gets a chance to fizzle out.
They have a lot of advantages in that they are low power, have built-in UPS, small, have a keyboard and display. The disadvantage is that you don't get a lot of storage options. I have some USB 3 attached storage, but I wanted something a little different this time.
Most recently my RPI 4 8GB serving as media server got replaced by my girlfriend's Asus laptop. She knocked it off a table and the corner of the display got smashed in. Now it is running Linux and Kodi, among other things. Can barely see it sitting there under the TV.
A bit better at 4K out then RPI 4, but not by a whole lot. Don't do any transcoding on it, though.
My "Big Servers" for my home lab are 3 ancient HP 1U rack servers. Built a small horizontal rack so they "hang down" instead of sideways and they take up almost no room. They run some high efficiency Xeon processors so I can get away with running them off of one household outlet. I only leave one on and spin up the other two when I need the extra capacity for some project.
Dual socket 8 cores and 72GB of RAM each.
Figured out that if you go and look at Vmware hardware support and get a server that is _just_ one generation older then they are willing to support than these things are dirt cheap on the used market.
If they can't run the latest version of Vmware then nobody wants them. Linux don't care, of course.
I think I spent a total of 600 dollars for all 3 servers, but it could of been a lot less. It's been a while.
I hate dealing with server-grade hardware because the lights out stuff is insecure (100% get one with a dedicated LO network port), dealing with firmware configuration is tedious, take forever to boot up, and dealing with the raid controllers is misery. But it's the cheapest way to get a ton of capacity. Especially if you have a project that wants to use IPMI and such things.
Licensing requirements for VMs and lack of server hardware that can be put in datacenters are both obstacles to target Apple Silicon. Not insurmountable, you can build enclosures for a rack of Mac Studios & Mac Pros and orchestrate VMs with Hypervisor framework, but it's a large enough obstacle that major cloud providers don't offer it and that pushes large costs onto the CI providers, which then get pushed to the folks trying to build tooling that works on M1 Macs. That includes the open source ecosystem.
The alternative is cross compilation, either cross-arch or cross-platform or both. All of these options are brittle and complex.
Speaking from experience here, I've had to follow issue trackers for: programming language teams shipping parity for M1, major libraries (think of foundational libraries like GRPC's libraries for each language) not yet shipping binaries for M1, and developer tools and OSS teams producing binaries. All of this needs to come together, and it hasn't yet.
The whole ecosystem is straining under the weight of this complexity to target engineers' laptops.
https://aws.amazon.com/ec2/instance-types/mac/
You can choose from either x86 or M1 machines.
It's been nearly two years since the Apple Silicon kits first shipped, and the only instances you can purchase from AWS are metal instances, at approx. $10k/year/instance.
That's not accessible to open source/free software projects and not affordable or efficient to utilize for small projects and highly intermittent workloads like CI.
I wish MacOS licensing were available for Graviton and ARMv8 generic, that would be incredible. It isn't, and the costs to support Apple's platform is a substantial barrier for many projects.
Hobbyists? I spend $700/year to have access to a maker space. I spend that much on woodworking tools annually (if not more). Hobbyist doesn't mean free.
That said, Apple isn't going to sell chips these to anyone else, period. And everyone right now is hustling to match the same levels of efficiency in newer designs, and they'll get there sooner or later. So "Apple Servers" are a completely moot point, but power efficiency? Yes, it literally matters more than ever, across every spectrum of the industry, and it's why the M1 was something people cared about anyway.
It’s done to death that their web services are of dubious quality, and they have got better. However you don’t have to go very deep or use services for very long to find some very sharp edges.
If they did go into this business seriously they’d probably do this again. That kind of support doesn’t scale for consumer products.
…that it will not happen via an add-on “server” program.
But TBH it’s hard to see how servers could be much of a market for Apple. People use the cloud more than local (“departmental”) servers these days and the cloud back end belongs to Linux, typically with a provider on top (AWS etc). Apple’s strengths lie elsewhere.
I am aware of the command to stop apple devices writing these files to network shares, but regularly our mac users 'forget' to run this command after setting up new profiles or upgrading OSs etc.
and reboot.
Terrible naming on that one...
Also, these files provide a useful service to Mac users. You could also find a way to support them instead of fighting this.
I appreciate it may come across as a bit hostile towards the users, but in reality in a budget constrained environment where we cant make it perfect for everybody, we must make sacrifices to make the majority of users have a better experience. In my opinion mac users not having metadata of when a file was last modified pales in comparison to the rest of the business searches taking twice as long on samba shares.
Isn’t that what’s making searches slow, not that the files exist? Why do you suppose that this isn’t a problem for Macs too?
That being said, depending on your business and data structure this doesnt detract from the point. Also if you have a deep nested directory structure with no real files in it, would every level contain a .ds_store file?
There are also AppleDouble (._) files, which are one-per-file. These contain the file's extended metadata and resource fork. Deleting these _may_ cause data loss if there's anything important stored in the resource fork. A better option is to enable vfs_streams on your Samba server to allow storing the additional forks natively on your filesystem (e.g. as xattrs).
(If you're using a modern Windows file server, I believe the resource fork is automatically mapped to an NTFS alternate data stream.)
See:
https://www.samba.org/samba/docs/current/man-html/vfs_stream...
https://www.samba.org/samba/docs/current/man-html/vfs_stream...
I also object to my operating system running all these background processes on MY computer without me commanding it to, and it suggesting on its own that I do this or run that, again in absence of any command to do so. More and more, operating systems and applications are treating MY computer as a dumping ground and science experiment: for things it wants to do, instead of what I want it to do.
I should not have to go off and find a setting somewhere just to stop my operating system from doing things on its own I don’t ask for.
It's worth noting that almost all modern file systems support multiple streams. NTFS has alternate data streams, Ext4 has xattrs. Modern SMB and NFSv4 also both support this at the protocol level.
The problem arises when you're using Samba (without vfs_streams enabled), or you're writing to a legacy FAT filesystem, in which case you start getting the AppleDouble files - again, to prevent data loss.
What about linux that would have 27 in this case?
https://support.apple.com/en-us/HT208209
They already have a solution but they're not using it?
With respect to DS_Store files on shared network drives, that is not true. Such files provide utility to a single Mac user, whichever uploaded the DS_Store file to that directory last. This file is used to store user preferences, which breaks as soon as there are two or more users with different preferences. Simply put, it does not belong on shared network drives at all.
My home NAS file shares are used by only me, so I'd like it to be supported there. So at the very least it needs to be configurable per-share. And in that case, having the server admin add them to veto lists seems like the easiest solution.
EDIT: I have had people say this to me before about windows and thumbs.db. But I personally have not seen this in the wild. Maybe its what old versions of windows did and people are still remembering this?
https://serverfault.com/a/5567
Thumbs.db files are created on my windows 11 pc at least. They’re only created for files that have metadata that requires reading those files. Explorer likes to display the metadata (sometimes) for some folders that have a lot of media in it (pictures, music, videos, etc). If the thumbs.db file is missing, windows will partially read every media file on the server to show thumbnails, that obviously creates unnecessary load but it’s really a trade off that might not make sense for most.
Full ext2/3/4 read-write support. $39.95.
There's also fuse-ext2 via macFUSE but its doc still say: "Even though write support is available, please do not mount your filesystems with write support unless you have nothing to lose."
https://github.com/alperakcan/fuse-ext2
https://www.paragon-software.com/home/extfs-mac/
It also doesn't seem surprising to me that Apple itself wouldn't support ext2/3/4. It's something very few Mac users are going to care about. Why would Apple dedicate any resources to it?
FWIW, I've been using Linux and Macs for over two decades and I can only recall maybe a single time I even wanted to mount an ext formatted drive onto a Mac, and vice-versa (HFS onto Linux). For exchanging files via floppy disks or thumb drives, various flavors of FAT have been near universal for ages. CDs/DVDs/BluRay use their own filesystem. And FTP/NFS/SMB/AFP/HTTP/SCP/RCP have been around forever for exchanging over a network. (Before ssh/scp, there was unencrypted rlogin/rsh/rcp... I'm dating myself.)
- FAT32 can't go above 4GB files, can't do sparse files and can't store UNIX permissions at all
- NTFS can do large and sparse files, but can't deal with UNIX permissions, instead bringing NT ACLs (which no one else understands). exFAT used to be patent encumbered and has the same limitations that NTFS has, plus its support is nowhere near as battle-tested as NTFS is
- HFS/APFS are not supported at all under Windows and support in Linux is spotty at best
- ext4 has third-party read (and if you're risk-tolerant, write) support on Windows and Linux, but it's hacky or expensive
The situation, as it is, is an utter shame.
FWIW, the company I linked elsewhere provides full ext4 support products for both macOS and Windows.
Shoving data across operating systems using USB sticks or drives is a pretty standard thing to do, at least in the corporate world which is filled with slow internet connections (in related words, never underestimate the bandwidth of a truck filled with SD cards), and particularly when having to send data to clients or vendors.
The fact that there is no standard that works on all three major OS families and in the best case also supports encryption is maddening - macOS has Filevault, Windows has Bitlocker and Linux has LUKS, and neither of the three can understand any other.
Just ask an IT helpdesk in any major company how many questions they get a day by people who have to transfer files to a vendor or client, with encryption. Extra fun in anything involving creative work because most of that happens on Macs whereas most big-corp customers use Windows because of Active Directory and GPO. (And no, ZIP isn't an answer because Apple's built-in unarchiver routinely gobbles up when encountering password protected ZIPs)
> I don't even understand the use case for file permissions on an external drive.
Unlike Windows which conveys the fact that an executable is an executable by the file suffix .exe (or .cmd/.bat, and I think *.js was at least until Windows 7 associated with the JScript interpreter), Linux and macOS do so by using the execute permission on the file.
The use case would be to copy a Java application, let's use jadx, downloaded and extracted on a Mac, then copy it via an USB stick or a Windows CIFS file share on another Mac or a Linux machine, and have the command bin/jadx-gui still start up the UI when clicked upon. As soon as any non-Unix filesystem is in the copy path, it gets inevitably nasty.
There's a version of 7zip for windows, linux and Mac. It's supposedly safe for now. Until the next vulnerability is found I guess but that is the same for every option.
Bit of a downside that it is a third party option that would then have to be installed everywhere you want it rather than a native option though.
Latency is the issue! Once I was working at a museum and we needed to get these stupidly high detail 3D scans to the university. Considering the IT policy and available bandwidth it was quicker for us to put all the files on a single HDD and walk it over.
File permissions usually don't make sense on an external drive that's moving between computers, much less computers running a different OS.
I just don't see why it's surprising that Microsoft and Apple don't have built-in support for ext when FAT/exFAT support is good enough for 99% of interop use cases.
exFAT is better, but it is also quite recent development on Linux, since it was licensed and only in 2019 that Microsoft publicly said, they would not ask for any licensing and the specification became publicly available.
Both of them are quite unfriendly to flash memory, but exFAT is good enough.
Files up to 16 EiB, supported from Windows XP and up, Mac's with OSX and up, Linux had some bugs back in 2010 which prevented you from dealing with very large devices that were over 80% full.
Been doing it for over 10 years now, and it is still one of the best formats for universal access.
Really it would be nice if everyone could just natively read EXT3/4 out of the box - theres no reason NOT to add it. And even though Microsofts new file system still doesn't seem to have taken off (ReFS) it is quite good.
Mostly it doesn't. And in any case, Windows doesn't ship with native support for mounting ext either. Only recently by installing WSL can it do so. So a very tiny minority of Windows users care about it.
https://devblogs.microsoft.com/commandline/access-linux-file...
Very hard.
Instead there are 3rd party implementations. Those are independently maintained, serving directly the (comparatively small) target audience.
What I'd REALLY love is to be able to run all of my own services in place of iCloud. I realize that this stuff isn't free to develop and I'd happily pay for it. They've been moving further away from standard formats and protocols with no way for others to integrate in the same way and I'm curious if there are any legal anticompetitive actions on the table should they continue to do so.
Before anybody says it— I have business needs that favor using MacOS directly on Apple Hardware and as a result, iOS is a good choice. Moving from Apple would regularly cause me more grief than these things do. No, you don't know my use case better than me and I have exhaustively explored all alternatives both ancient and modern.
I'm not sure if that will work on iOS since iOS doesn't really have a filesystem - there are apps available on iOS for Syncthing and Obsidian, but I don't know if Obsidian will be able to access the Syncthing synced folder. But it works great for me on Linux PCs + Android. Much better than Nextcloud which is too complex and is a pain to administer over time. Syncthing is super simple to set up does file sync very well, much better than Nextcloud does. Radicale is also easy to set up and just works.
The nice thing about the groupware functionality in Apple Server is that it used all of these standard protocols so it was entirely interoperable with other devices AND it had a nice smooth administration experience.
At the moment, I just pay $15/mo for Cloudron which handles email, is decently smooth for administration though a little more disjointed between the apps than I'd prefer, and can "one click" deploy Radicale, NextCloud, Sogo, et. al. I used to administer servers but it's not what I do now and I have no interested in sinking non-work time into work-like tasks.
You can configure Samba to store resource forks using xattrs instead. See fruit:resource:
https://www.samba.org/samba/docs/current/man-html/vfs_fruit....
You can veto the .DS_Store files. The consequence to Mac users is that the Finder won't remember any display changes they make to windows that correspond to network folders.
https://www.samba.org/samba/docs/current/man-html/smb.conf.5...
(Note that "._*" prevents HFS+/APFS extended attributes from being stored on the SMB drive.)
Good. Why would one user's view preferences have an effect on another user? That is what happens if .DS_Store files are left on the server.
If you have readonly access to the folders you can't persist a change to the layout.
Or at least I hope.
At the end of the month it sums up to a complete employee pay check in addition to the employee pay check.
Sure, there are some large companies that the total cost of subscription services would add up to the equivalent of 1 more employees paycheck but is that really so crazy? For what they get out of it in terms of not having to pay multiple IT people it seems like a pretty decent deal. Someone has to maintain/update/support/etc the tools, it's not free.
Source: https://support.apple.com/en-us/HT213238
No mention of Time Machine.
No mention of Caching Server
No mention of File Server.
Three things that happens to be what I thought important apart from MDM, gets no mentions and I assume no use at all? Caching Server used to cache iCloud content as well. Never tried it with iCloud Photos though so I am not sure if it is a backup solution if Apple ever log you out of APPLE ID.
This basically ends the dreams of Apple ever selling a Time Capsule with iOS backup and NAS functions.
> The most popular server features—Caching Server, File Sharing Server, and Time Machine Server are bundled with every installation of macOS High Sierra and later, so that even more customers have access to these essential services at no extra cost.
For some strange reason their sentence wasn't immediately obvious to me these services are staying.
Not sure about iCloud, I setup for updates at a school.
File servers are commodity, legacy tech, I don’t see a need for a first party solution.
https://www.cyrusimap.org/imap/concepts/features/event-notif...
It's too bad that, assuming that documentation is still up to date, Apple doesn't allow smaller IMAP mail providers to integrate with the push notification service.
https://github.com/freswa/dovecot-xaps-daemon
https://github.com/freswa/dovecot-xaps-plugin
We would love to be able to send push notification to native apple Mail app for our users but we never wrapped our head around the whole certificate thing. (And Apple was... unhelpful).
Would love to have support on this! Are you available? Let me know (nicolo.benigni@qboxmail.com)
ABE and ABM both use the same technology so it's pretty much exactly how Apple intends it to be used. There used to be someone hosting a shared CA for this, but it because cumbersome to deal with third parties not loading the CRL and not using OSCP.
E.g., "I'll be able to test that code change in a bit, I got Jamf'd pretty hard this morning and now my build is broken..."
It's just a management framework - Do developers who use windows say "getting sccm'd" or "getting Intuned" when SCCM/Intune does management actions on their devices?
> I got Jamf'd pretty hard this morning and now my build is broken...
If the IT department is pushing stuff out through the management framework on the devices that is "breaking builds" - they are doing device management wrong.
And if the IT department is pushing out Config Profiles which break things people use every day, thats more a people problem than technology/Jamf problem. (as in the IT dept are morons)
Maybe software like Jamf isn't inherently a problem, but it can encourage IT departments with that kind of culture to do more things to irritate developers (like automatically run buggy scripts on developer laptops without any notice or easy visibility into what those scripts are and what they do), which without that kind of management software available to them, they would have been less likely to try to do. Developers are more likely to have configured their environment in custom ways which will cause IT's buggy script to play up and cause problems
I used to work for Oracle, and while I was there Oracle had a whole separate IT department just for the engineering/R&D org, while the main IT department serviced the rest of the business (sales/support/finance/legal/HR/etc). I don't know whether Oracle still uses that setup, but (if the company is large enough to sustain it) that could be an approach to avoid some of this
When IT, Security, and compliance are stepping over each other with GPO, that’s when you really suffer.
My favorite was when some jackass deployed something via GPO that dropped a script in user space that triggered a serious Crowdstrike alert.
Lots of random 99% CPU... intervals. Although that hasn't happened lately.
Anything involving tons of files (so, Node) takes like 2-4x as long as it does on a Mac that's not loaded down like this. It's painful.
Not sure how much of this is due to JAMF and how much of this is due to the other wonderful apps.
Ubisoft Massive does have the phrase "got GPO'd" if that helps.
In my timeline the Server macOS died somewhere around trashcan Macs introduction or maybe even earlier.
Works very well.