Are these windows-specific issues?
If so, would not it be better to drop this dying platform and focus developer time on what actually matters? (Nothing against Vadim personally 
)
1 Like
Consider them a regularization framework: if software works even on Windows, then it will operate even more efficiently everywhere else 
4 Likes
Yes,
Not so dying:
Most of those changes are not investment into software robustness but a Sisyphean busywork fighting windows specific windmills, that produce fragile workarounds that get in the way of maintainability of the whole product.
Makes no sense for a declining ecosystem. Servers don’t run on windows anymore. It’s now solely ads deliveryinto the consumer throats machinery and legacy systems crutches: both captive audiences that can and do tolerate a lot of abuse.
Why is storj wasting time with it — no idea.
2 Likes
v1.142.7 Windows: 114,167 Non-Windows: 129,073 Windows %: 46.94
v1.141.2 Windows: 238,152 Non-Windows: 225,565 Windows %: 51.36
v1.139.6 Windows: 80,443 Non-Windows: 204,315 Windows %: 28.25
v1.137.5 Windows: 93,679 Non-Windows: 235,497 Windows %: 28.46
v1.136.4 Windows: 82,416 Non-Windows: 171,794 Windows %: 32.42
v1.135.5 Windows: 53,637 Non-Windows: 121,332 Windows %: 30.66
Depressing. Who are those people? Why are they doing it to themselves?
I’m sure if we drop support – the majority those folks will migrate to a better OSes, and a huge load off of developers shoulders.
Important questions: – how does reliability, turnover, tenure, and audit/online/suspension metrics looks like in widows vs non-windows cohort?
1 Like
please consider that your way is not only one right way. for example I dont like your crap linux. Most things that I can do easy on windows, need to do manually on linux.
1 Like
I despise linux as well. Linus is almost always a compromise, and it’s now popular for the same reason windows became popular in the nineties, that have little to do with its technical merit.
Does not matter, this is the reality, linux won. Might as well embrace it. (I don’t – I stick to FreeBSD, and I’m in the same minority, but I’m willing to die on that hill).
This is a matter of habit, not virtue of a platform. Personally, I have used windows for far longer than any other OS combined, and I still despise it. And yes, I agree, linux is way worse – but it’s not its fault really. Linux is a kernel, the rest of haphazard zoo of user spaces that are build with no restraint or coherence. The result is universally garbage. I would not use it as a desktop platform. if you want coherent system – FreeBSD is the way to go. It’s the whole system, not just a kernel. Or macOS, if you also want hardware. Knowledge is mostly transferable.
But we are talking about services here.. and in services linux is king. Including due to design choices that enabled network scalability… somewhat. Netflix is still using FreeBSD for their CDN. So there is that.
I don’t know what systems are running nodes, but who runs windows, needs to, because:
- many business specific programs are only developed for Windows;
- gaming is better on Windows, video drivers are better on windows (at least it used to be);
- GUI is way better on Windows than Linux, and, for many, CLI is out of discussion because you spend your time way more efficient on GUI, instead of trying to remember/guess/read manual pages of commands in a CLI enviroment.
- in general, programs are way more developed for Windows than Linux. Many devs make the Win version first, and than, maybe, a Linux one.
- I never felt the need to complicate my life with Linux, until the need/desire to run storagenodes; and I started with Windows nodes.
2 Likes
Unfortunatelly, you see a people’s choice. freeBSD has 8 downloads, despite its superiority.
Please, do not suggest to drop support of a minority of your choice. Because it will be not Windows… 
I still protecting to keep freeBSD versions, because freeBSD is much better in many aspects. However, we should accept the reality, even Darwin has much, much less than Linux, I wouldn’t say about freeBSD.. And not about Windows. Currently it’s N1.
Doesn’t matter that most of the team are on mac’s and Linux, except some, like me, because I need to have all versions.
This is the capture audience I alluded to above.
Right. We are not taking about gaming rig. We are talking about servers with unused capacity that has space to run stroagenode. Gaming rigs are out of scope.
Ye…ish.
False. And matter of preference at best. CLI is always, way, way faster – but this is also irrelevant. We are talking about servers here.
False. You are conflating user apps with server software. Most server platforms are linux. Almost none are windows. See the link I posted above. Less that 10%, and declining fast.
So you pay for windows server, just to … why exaclty?! Makes no sense to me.
It’s fine if windows support was free. Clearly, it is not. So why suffer forever instead of switching once?
Again, not talking about a desktop. Use what you like. But windows on a server that requires uptime is simply goofy.
Look it like this: if it was only for Linux, the majority of current Win users maybe would have never tried to run a storagenode; but with a Win version available, they had the chance to, and doing so, they wanted to expand (myself), run better or cheaper nodes, and switched to Linux nodes. And Storj brought Linux in their lives.
2 Likes
If freeBSD is dropped, I’ll delete nodes once the current version expires. No hard feeling.
Darwin should not be in the picture at all. There are no macOS servers anymore. It’s a desktop OS. Xserve is dead long time ago.
It’s clearly not N1, it’s’ 30-50%. So half.
What I want to know is reliability and turnover of windows vs non-windows nodes. Can you share these stats? There are more costs of supporting windows than code maintenance – most windows hosts I suspect are crappy gaming rigs, where reliability is an afterthought that never materialized. And NTFS is not suitable for anything serious. Microsoft can’t design filesystems, I hope it’s obvious by now? ReFS, anyone ?
I don’t have this stat. If consider the amount, then Windows nodes are more reliable (because of an amount, of course…).
However, since we do not use Windows in a dev processes with a few exceptions, it’s likely not a picture to show.
I’m agree. But even on my own experience, I not always can run even Linux (due to HW issues).
Accoridng to that gihub link, windows is till less than half. Did you see that post above?
And I agree, number of downloads maybe be bad metric – maybe a single machine sucks at downloading and redownload 600 times the same file trowing all stats 
I hope storj has telemetry of node OSes, and someone shall be tracking this.
The despicable docker exist, no? Or even WSL2 and what not? Windows users can still run linux software, especially self-contained go applications like storage nodes. So technically you can still drop windows build pipeline and still let windows users participate, until they eventually migrate to better systems.
Even I could run linux in bhyve and continue hosting nodes. So my earlier statement was not thought out. I don’t have a problem with that. Feel free to drop FreeBSD. I do run one of the VMs for a service that is linux only. Not a big deal. But the foundation remains solid – filesystem, networking, and services are freebsd. And uptime is magnificent as a result. And no support burden from me to storj. Except to forum moderatos…
1 Like
Yes, I see. The current version is 1.142.7
freeBSD: 41
Linux amd64: 120892
Linux arm: 666
Linux arm64: 7481
Windows amd64: 272132
What I mean, that Windows is N1 vs all others.
Yes and I tried to migrate to a native Linux for my existing hardware, I had failed.
I know, the HW is a crap. However, it worked under Windows since 2014.
Where do you get his from?
From https://api.github.com/repos/storj/storj/releases/tags/v1.142.7
storagenode_windows_amd64: 114160
\
from here: https://api.github.com/repos/storj/storj/releases/assets/320212314
This is the actual number of nodes, which are exists 
Because some using hacks to update themselves.
Ok, I’m very confused about why two URLs show different counts, but fine, this is all a proxy to the actual numbers anyway.
Actual numbers are known to a satellite, if it tracks OSes as part of telemetry. So that should be the authoritiave source of truth, not gihub downloads that don’t even agree with each other.
These figures are huge when there is 32k nodes in operation.
What about DoubleSpace?