I've said for a long time that Linux & open source is kind of a paradox. It goes everywhere, it speaks every protocol. But as a client, as an end. The whole task of coordinating, of groupwareing, of bringing together networks: that's all much harder, much more to be defined.
Making the many systems work together, having directory infrastructure: that stuff is amazing. For years I assumed that someday I'd be running FreeIPA or some Windows compatible directory service, but it sort of feels like maybe some OpenID type world might possibly be gel'ing into place.
Appreciate that! Simple login and access was certainly the hardest requirement to hit, but it can be the difference between people using something and not
And I agree with the feeling that open source is everywhere, up until a regular user picks up something. I think part of the paradox you mention is that every project is trying to work on their own thing, which is great, but also means there isn't a single entity pushing it all in one direction
But that doesn't mean we can't get to nice user experiences. Just in the self-hosting space, things have gotten way more usable in the last 5 years, both from a setup and usage perspective
It's not supposed to be. You put in time, use your brain to understand the system. Even a non-techie can easily understand OIDC and Oauth2, it's not that hard
As a techie, experienced in security, reading the OIDC spec... there are definitely some things I don't understand in there. I'm not sure the authors even understand what's going on.
On 2023-12-15 they published an update to OpenID Connect Core 1.0, called "errata set 2". Previously it said to verify an ID token in a token response, the client needs to
> * If the ID Token contains multiple audiences, the Client SHOULD verify that an azp Claim is present.
> * If an azp (authorized party) Claim is present, the Client SHOULD verify that its client_id is the Claim Value.
The new version is quite different. Now it says
> * If the implementation is using extensions (which are beyond the scope of this specification) that result in the azp (authorized party) Claim being present, it SHOULD validate the azp value as specified by those extensions.
> * This validation MAY include that when an azp (authorized party) Claim is present, the Client SHOULD verify that its client_id is the Claim Value.
So core parts of the security of the ID Token are being changed in errata updates. What was the old purpose of azp? What is the new purpose of azp? Hard to tell. Did all the OIDC implementations in existence change to follow the new errata update (which didn't update the version number)? I doubt it.
Or how about a more fundamental question: Why does the ID Token have a signature? What attack does that signature prevent? What use cases does the signature allow? The spec doesn't explain that.
The ID Token can be passed from the Identity Provider to the Relying Party (RP) in a few ways.
When `response_mode=id_token` is used, the ID Token can be passed in the front channel directly to the RP during a browser redirect. Since the ID Token is coming from the browser, it must be signed to ensure that a malicious actor can't tamper with it. Otherwise an actor could swap out a `sub` or an `email` claim and the RP would be none the wiser.
The ID Token can also be returned from the `/token` endpoint after exchanging an authorization code. Since the `/token` endpoint is a back channel call over HTTPS, the ID Token doesn't necessarily need to be signed here to avoid tampering. The RP can trust that TLS gets the job done. However, there are substantial benefits to having it be signed:
- If ID tokens were only signed sometimes, we'd have two different standards for how to construct and handle an ID Token, which would be quite confusing.
- Signed ID Tokens can be passed around to assert identity to other entities within the system. For example, there are some promising draft specifications that explore exchanging ID Tokens for access tokens in other systems. This is only possible because the ID Token cannot be tampered with.
>Since the ID Token is coming from the browser, it must be signed to ensure that a malicious actor can't tamper with it. Otherwise an actor could swap out a `sub` or an `email` claim and the RP would be none the wiser.
Are you referring to a login CSRF attack? Where an attacker causes a victim to visit a URL that forces the victim to log in to an account of the attacker's choosing? A signature on the token doesn't solve that AFAICT. The reason a token signature doesn't solve that is that an attacker can take the attacker's own ID Token that has a valid signature, and do the attack with that ID Token, forcing the victim to log in to the attacker's account.
Yes, the token signature does reduce the attacker's ability. Without a token signature, the attacker could force the victim to log in to any account. With a token signature the attacker can only force the victim to log in to the attacker's account. But this isn't a full solution. Some other anti-CSRF mechanism needs to exist. Once that anti-CSRF mechanism exists, the token signature is no longer useful.
Adding a signature to the token doesn't make sense to me as a login CSRF protection mechanism. If the designer's goal was to protect against login CSRF, the signature should be over something else, not just the token, and not part of the token. Then the token itself doesn't need the complexity of a signature (and the authorization code flow doesn't need the complexity of a signature), just the redirect flow needs it. Or even simpler, the designers could have mandated the authorization code flow, and not need any signature at all.
Or maybe you're referring to a completely different attack, where there isn't a victim user, just an attacker wanting to log in as someone else in the attacker's own browser. In that case, the signature doesn't solve that attack, because since everything happens in the attacker's browser, the attacker can modify the js locally to disable signature verification.
>- Signed ID Tokens can be passed around to assert identity to other entities within the system. For example, there are some promising draft specifications that explore exchanging ID Tokens for access tokens in other systems. This is only possible because the ID Token cannot be tampered with.
Yes, but the spec doesn't mention this. Just from reading the spec, a reader can wonder, "why is there all this unexplained complexity for no apparent reason?"
> Did all the OIDC implementations in existence change to follow the new errata update (which didn't update the version number)?
I mean, both the old and new version (at least, the parts quoted upthread) are exclusively SHOULD and MAY with no MUST, so (assuming, for the SHOULDs, the implementer had what they felt was sufficiently good reason) literally any behavior is possible while following the spec.
It's not really that hard to be honest. If you are not dead set on specific services, but make sso compatability the main selection metric for the services it's very feasible and not that difficult. I had very little experience when I set up my self hosted system and was set up very quickly using caddy and authentik. Alternatively yunohost is a very easy to use distribution that sets up everything using SSO.
Agreed. I use Cloudflare Zero Trust for SSO authentication (I use Github for my self and Google/Apple for family) and Cloudflare Tunnels to expose homelab services behind NAT. It took an afternoon to set up the first service and adding services via IaC like terraform is easy.
The only time consuming thing since then has been figuring out how to use the Cloudflare auth header to authenticate with each individual app (and many don’t have multiuser capability so it’s not a big deal).
Yuno-host doing SSO is something I deeply deeply respect & love about it! Yes!
I haven't pulled it down yet to find out just how extensive the SSO is. My understanding is that it's not totally universal... But that it is like ~50%, which is super not bad given how many packages it has!
Hey y'all, I know getting a setup that feels "right" can be a process. We all have different goals, tech preferences, etc.
I wanted to a share my blog post walking through how I finally built a setup that I can just be happy with and use. It goes over my goals, requirements, tech choices, layout, and some specific problems I've resolved.
Where I've landed of course isn't where everyone else will, but I hope it can serve as a good reference. I’ve really benefited from the content and software folks have freely shared, and hope I can continue that and help others.
How are you finding Nix for the homelab to be? Every time I try it I just end up confused, maybe next time will be the charm.
The reason I ask is I homelab “hardcore”; i.e. I have a 25U rack and I run a small Kubernetes cluster and ceph via Talos Linux.
Due to various reasons, including me running k8s in the lab for about 7 years now, I’ve been itching to change and consolidate and simplify, and every time i think about my requirements I somehow end up where you did: Nix and ZFS.
All those services and problems are very very familiar to me, feel free to ask me questions back btw.
I certainly didn't take to Nix the first few times I looked at it. The language itself is unusual and the error messages leave much to be desired. And the split around Flakes just complicates things further (though I do recommend using them, once you set it up, it's simple and the added reproducibility gives nice peace of mind)
But once I fully understood how it's features really make it easy for you to recover from mistakes and how useful the package options available from nixpkgs are, I decided it was time to sink in and figure it out. Looking at other folks nix config on GitHub (especially for specific services you're wanting to use) is incredibly helpful (mine is also linked in the post)
I certainly don't consider myself to be a nix expert, but the nice thing is you can do most things by using other examples and modifying them till you feel good about it. Then overtime you just get more familiar with and just grow your skill
Oh man, having a 25U rack sounds really fun. I have a moderate size cabinet I keep my server, desktop, a UPS, 10Gig switch, and my little fanless Home Assistant box. What's yours look like?
I should add it to the article, but one of my anti-requirements was anything in the realm of high availability. It's neat tech to play with, but I can deal with downtime for most things if the trade off is everything being much simpler. I've played a little bit with Kubernetes at work, but that is a whole ecosystem I've yet to tackle
>The language itself is unusual and the error messages leave much to be desired. And the split around Flakes just complicates things further
Those are my chief complaints as well, actually. I never quite got to the point where I grasped how all the bits fit together. I understand the DSL (though the errors are cryptic as you said) and the flakes seemed recommended by everyone yet felt like an addon that was forgotten about (you needed to turn them on through some experimental flag IIRC?).
I'll give it another shot some day, maybe it'll finally make sense.
>Oh man, having a 25U rack sounds really fun. I have a moderate size cabinet I keep my server, desktop, a UPS, 10Gig switch, and my little fanless Home Assistant box. What's yours look like?
* 2 UPSes (one for networking one for compute + storage)
* a JBOD with about 400TB raw in ZFS RAID10
* a little intertech case with a supermicro board running TrueNAS (that connects to the JBOD)
* 3 to 6 NUCs depending on the usage, all running Talos, rook-ceph cluster on the NVMEs, all NUCs have a Sonnet Solo 10G Thunderbolt NIC
* 10 Gig unifi networking and a UDM Pro
* misc other stuff like a zima blade, a pikvm, shelves, fans, ISP modem, etc
I'm not necessarily thinking about downsizing but the NUCs have been acting up and I've gotten tired of replacing them or their drives so I thought I'd maybe build a new machine to rule them all in terms of compute and if I only want one host then k8s starts making less sense. Mini PCs are fine if you don't push them to the brim like I do.
I'm a professional k8s engineer I guess, so on the software side most of this comes naturally at this point.
For certain definitions of the word “fun,” yes. I have a 35U (I don’t need that many slots, but at the time I did need it tall enough that my kids couldn’t reach the top, where I put the keys), with:
* 3x Dell R620
* 2x Supermicro (one X9, one X11)
* 1x APC UPS w/ external battery
* Unifi UDM Pro
* Unifi Enterprise 24-port switch
The Dells have Samsung PM863 NVMe drives which are used by Ceph (managed by Proxmox), with traffic sent over an Infiniband mesh network via Mellanox ConnectX3-Pro.
The Dells run K3OS in a VM, which is a dead project. Big mistake there.
The Supermicros have various spinners, and are in a ZFS pool. One of them is technically a backup that should power up daily to ingest snapshots, then power off, but there’s been some issue preventing that, so…
It was all very fun to set up, and has been eminently reliable, but it’s a bit much. While you can in fact make R620s relatively quiet, they’re still 1U, and those little 40mm fans are gonna whine. It’s background noise to me, but guests definitely mention it if we’re close to the rack.
Also, I’m now in the uncomfortable position of being stuck on Proxmox 7, because v8 (or more accurately, the underlying Debian release) dropped support for my HBAs, so the NAS would be dead in the water. I mean, I could compile my own kernel, or perhaps leverage DKMS, but that somewhat defeats the purpose of having a nice AIO like Proxmox. Similarly, my choose of K3OS means at some point I need to spend the time to rip everything out and start over with Talos.
Or - just maybe - I’ve done enough playing, and I should simply buy a JBOD chassis and a relatively new and quiet server (4U under light load means you can get away with much quieter fans), and just run stuff in Docker or gasp systemd. Or, hell, single-node K8s. My point is that it is fun, but eventually your day job becomes exhausting and you tire of troubleshooting all the various jank accumulating at home, and you stop caring about most of it.
I turned my supermicro 847 X10DRI+ into a JBOD by ripping out the mobo, the fans, and building a 140mm fan wall (zip ties and later 3D printed). I can highly recommend the move, it's dead silent and I have drive temps between 33 and 45C with 20 to 25 drives (I honestly don't know the total at this point).
As for your OS issues I also used to run Proxmox with Ceph (btw did you know you can use Proxmox's ceph with rook-ceph so you don't need 2 layers of storage?) including for the router and NAS, but I gave it up due to unwarranted complexity and went bare metal.
Don't know what would fit your particular use case but I can say this: I'm very happy I made a separate box with a supermicro X11 and a JBOD besides it, I can recommend this too; what benefit is there really to virualizing a NAS?
Regarding K3OS you're in luck - kubernetes manifests (you've got GitOps, right?!) are so portable you can just rebuild on a new OS. Give Talos Linux a spin. Again really think about why is it that you're virtualizing here too; maybe you genuinely need it, maybe not.
Not sure I follow here; why would I want Rook involved? I generally don’t want my orchestration layer - which is also consuming storage - to be involved with the management of said storage.
> Virtualizing vs. bare-metal
It’s partially to make upgrades via new base images easier. I bake images with Packer + Ansible, and so can push a new one out quite easily. The other part is that my NAS consumes very little compute resources, since it isn’t hosting any apps, so it would be a waste to run it on bare metal. Tbf I could run everything consuming the disk storage directly on it, but I had shied away from that initially and it stuck.
> GitOps
I have Helm templates, no ArgoCD. It’s been a TODO for me for quite some time. Same with Talos (I actually do have it running in parallel right now, just not hosting anything). My issue is that I am obsessed with getting things perfect, and for me that means bootstrapping from a bare VM to ArgoCD ingesting manifests and spinning up pods, all automated. I know this is possible, as I’ve seen it done, but I rarely have the time or energy to pursue it after work. I should probably get over myself and just manually install stuff so it’s functional.
>What does your persistent storage layer look like on Talos?
Well, for its own storage: it's an immutable OS that you can configure via a single YAML file, it automatically provisions appropriate partitions for you, or you can even install the ZFS extension and have it use ZFS (no zfs on root though).
For application/data storage there's a myriad of options to choose from[0]; after going back and forth a few times years ago with Longhorn and other solutions, I ended up at rook-ceph for PVCs and I've been using it for many years without any issues. If you don't have 10gig networking you can even do iSCSI from another host (or nvmeof via democratic-csi but that's quite esoteric).
>How have you found it's hardware stability over the long term?
It's Linux so pretty good! No complaints and everything just works. If something is down it's always me misconfiguring or a hardware failure.
It was years ago but I recall high CPU usage being an issue in particular.
In general it's just not as battle tested as ceph and I needed something more bulletproof.
However I will say this: I'm sure that issue with the CPU usage was fixed (I was watching the GitHub issue) and you might not need your distributed FS to be CERN ready for your lab; AND the UI and built-in backups Longhorn offers are great for beginners so I'd suggest giving it a try if you don't already know you want ceph or OpenEBS Mayastor for the performance and so on.
Honestly, I personally like the combination of keepalived+docker(swarm if needed)+rsync for syncing config files. keepalived uses VRRP, which creates a floating IP. It's extremely lightweight and works like a charm. You won't even notice the downtime, the switch to another server IP is instant
Did you come across or consider using coolify at any point? I've been using it for over a year and quite enjoyed it for it's Heroku type ease of use and auto deployments from GitHub.
I have dabbled before with FreeIPA and other VMs on a Debian host with ZFS. For simplicity, I switched to running Seafile with encrypted libraries on a VPS and back that up to a local server via ZFS send/receive. That local server switches itself on every night, updates, syncs and then goes into sleep again.
For additional resiliency, I'm thinking of switching to ZFS on Linux desktop (currently Fedora), fully encrypted except for Steam. Then sync that every hour or so to another drive in the same machine, and sync less frequently to a local server. Since the dataset is already encrypted, I can either sync to an external drive or some cloud service. Another reason to do it like this is that storing a full photo archive within Seafile on a VPS is too costly.
I appreciated the in depth look and while some ideas from your setup will take more time to implement, I just added flame for the dashboard and see how it fares with family.
Hi! Really excited by your work! I'm working on a similar project built on NixOS and curious what you thing.
My goal is to have a small nearly zero-conf apple-device-like box that anyone can install by just plugging it into their modem then going through a web-based installation. It's still very nascent but I'm already running it at home. It is a hybrid router (think OPNSense/PFSense) + app server (nextcloud, synology, yunohost etc). All config is handled through a single Nix module. It automatically configures dynamic DNS, Letsencrypt TLS certs, and subdomains for each app. It's got built in ad blocking and headscale.
I'm working on SSO at the moment. I'll take a look at your work and maybe steal some ideas.
The project is currently self-hosted in my closet:
Sometimes when I think about my home network, I think about it in terms of what will happen when I die and what I will be inflicting on my family as the ridiculous setups stop working. Or like, how much it would cost a police forensics team to try to make any sense of it.
I think "home labbing" fulfils much the same urge / need as the old guys (I hate to say it but very much mostly guys) met by creating hugely detailed scale model railways in their basement. I don't mean that in a particularly derogatory way, I just think some people have a deep need for pocket worlds they can control absolutely.
I ahve exactly the same thoughts and I wrote a document to be used in case I die.
Part one is money and where the important papers are.
Part twonis hiw to dulb down my home. How to remove the smart switches (how to wire back the traditionnal switches). How to mive self hosted key services to the cloud (bitwarden, mostly) and what to pay for (domain and mail). Hiw to remove the access point and go back to the isp box.
My wife is not supportive of the smart stuff but now that she knows she can dumb it down she is fine. Honestly she does not realize what strp back the lack of all this stuff will be. But at least it won't be my problem anymore:)
I have our family pictures on a RAID 1 array in my home lab. Every night they are rsynced to an external drive on a little computer at my in-laws. Both as a backup, and as an "if something happens to me" easy access. My wife doesn't have any interest in tech, so I wanted to make accessing it "just in case" as straightforward as possible. I told her that that is where all the photos are, and that it's just a USB drive she can connect to her laptop in case something happens.
Just ignore the useless threat scenario of someone stealing your physical disks to obtain your precious family data and you will be fine. In other words, just store all the photos and important documents in clear, some written down instructions and you should good to go.
I'm more worried by home automation in my case ^^;
The chance of someone breaking in to steal your sensitive files is next to nil I agree.
The chance of someone breaking into your house is sadly much more likely, and them choosing to take any computers they see is almost a certainty at that point.
Your drives are unencrypted. What's your next step if you come home tonight and find the house ransacked and the server gone?
It can be a side-effect. Francis Ford Coppola lost all his family photos when his PC was stolen in a burglary while in Argentina to shoot “Tetro”. Of course, he didn’t have backups.
> Your drives are unencrypted. What's your next step if you come home tonight and find the house ransacked and the server gone?
My drives are encrypted and so are my backups (with backups everywhere). But they're symmetrically encrypted with a password. The backup procedure contains a step verifying that decryption works.
Family knows the password: password is stored at different places on laminated paper (friends and family) but not alongside the backups.
Decryption of the backups is one command at the CLI (both brother and wife knows how to use a CLI and soon the kid shall too: already dabbled with it).
The one command is explained alongside the password, on the same laminated paper as the backups.
Yup I did really think this out, including rehearsals where I, literally, fake my own death (I fake a heart attack) in front of my brother and wife and I have to shut the fuck up while they open a CLI, hook up one of the backup hard disk and decrypt the backups.
Once a year we rehearse.
That way they are confident they can restore the backups. I know they can and I don't need reassuring, but they do (well less and less because know they began realizing I really thought this out).
> The chance of someone breaking into your house is sadly much more likely, and them choosing to take any computers they see is almost a certainty at that point.
Got a house break in years ago, they stole no computers.
> What's your next step if you come home tonight and find the house ransacked and the server gone?
Go to the bank, take of one my backup hard drive. Buy a computer, reinstall Proxmox, a VM, Docker CE, redeploy my infra. They still don't have the Yubikeys on my keychain. They still don't have what's on my phone.
Don't think some people here didn't plan for death / theft / etc.
I think a much more likely scenario is an unencrypted drive fails and then what? Do you send it to the landfill unencrypted? Or do you have some process to physically destroy it? Encryption means you can just toss it and feel reasonably confident the data isn't coming back to haunt you.
You should see the hilaribad basis given in affidavits for search warrants that get rubber stamped by judges.
There is no burden of proof and no consequence for perjury. 100% of the search or seizure warrants I have read have had obvious perjury in them.
I encrypt my data at rest not because I fear a burglar breaking in, but because I fear the FBI coming in the front door on some
trumped up bullshit. Everyone has a right to privacy, even (and perhaps especially) if they are doing nothing wrong.
I’ve read too many stories of writers and activists getting bogus warrants and charges and arrests thrown at them to inconvenience and harass them to ever have a single unencrypted disk in my house.
I think planning for what happens once you aren't there to manage the setup (whether it be a vacation, hospital stay, or death) is important. It's not something I built specifically to make easy and I should think more on it
The most important thing is to be able to get important data off of it and have access to credentials that facilitate that. You could setup something like Nextcloud to always sync important data onto other people's devices, so make part of that easier
But I think another important aspect is making folks invested in the services. I don't expect my partner to care about or use most of them, but she does know as much as I do about using and automating Home Assistant (the little we've done). Things like that should keep working because of how core they can become to living our lives. It being a separate "appliance" and not a VM will also help manage that
But also that's a lot of hope and guessing. I think sitting down with whoever might be left with it and putting together a detailed plan is critical to any of that being successful
I have given this a lot of thought. I assume the nas and its docker services won't boot starting everything up for someone else. My offsite encrypted backup is probably not recoverable without hiring someone. So:
- I have an ntfs formatted external USB drive to which cron copies over a snapshot of changed daily into a new folder. Stuff like paperless, flat file copy of seafile libraries. The size of that stuff is small <50gb, duplication is cheap. In event of death or dismemberment... that drive needs to be plugged into another machine. There are also seafile whole library copies on our various laptops without the iterative changes. Sync breaks... keep using your laptop.
- I've been meaning to put a small pc/rpi at a friend's place/work with a similar hard drive.
- the email domain is renewed for a decade and is hosted on iCloud for ease of renewal. Although I am not impressed that it bounces emails when storage is full from family member photos which happens regularly so may switch back to migadu.
This applies to so many other things. Who in your house does the taxes? If it's you, would your SO be able to pick up the slack in the event of your death? Can they access all the accounts? Do they even know what all the accounts are? I keep telling myself I need to put together a "what to do if I'm dead" Google doc, but haven't gotten around to it.
I pay $3/mo or whatever for Bitwarden family. It’s wonderful. My wife and I can access all our passwords (and OTP codes!) in one spot. I grouped passwords into folders like “Health” and “Finances”. It has taken us far.
1Password is amazing for this, IMO. My spouse and I have been using 1Password together for more than a decade. One of the first things I set up is a "AAA Read Me First" note with links to a bunch of other notes and documents, including our estate planning stuff.
The biggest thing that makes me stick with 1Password, despite the semi-recent VC shenanigans, is the fact that if for some reason we fall behind on billing (for example, because the credit card got cancelled because I died) the account goes into read only mode forever. As long as 1P is a going concern the data we choose to put there is safe from the biggest risk in our threat model.
Just a year before my dad's stroke, my parents documented every account, password, service they had; it was incredibly helpful after he passed with all the stress and chaos
There is a dead man’s switch service [1] which can send an email if you die. In theory if you self host you could trigger something when their email arrives to an inbox you control.
I’ve been thinking of making a version of this that does a webhook but it doesn’t offer a huge amount of value over the email method.
Is the dead man's switch necessary? Unless your homelab contains secrets you don't want revealed until after your death, I'd just put this in a Google doc.
Depends how convoluted your setup is. For some use cases releasing the location of passwords physically written down might help matters or maybe trigger a process to export all data and upload somewhere - somehow.
Seeing some of the discussions around home labs with server racks and k8s doesn’t fill me with confidence that for a majority of use cases a family member would be able to get the data if needed.
I'm glad to see this comment here. People build these projects for family and friends - which is great - and encourage their use, without considering what happens if the only sysadmin suddenly dies. You wouldn't let one person at work hold all of the keys, so the same should be true for your homelab.
While I haven't given all of my keys to my family, there's a clear route for them to get them, and written instructions how to do so. Along with an overview of the setup and a list of friends and colleagues they can turn to, this is enough for them to get access to everything and then decide if they want to carry on using it, or migrate the data somewhere else.
To be frank, if you die, isn't it much more likely your friends and family will just stop using your homelab setup? They'll switch back from Jellyfin to Netflix, replace the smart light bulbs with regular ones, etc.
Unifi network; small proxmox vms for core services; big truenas box for movies, storage, "apps ecosystem" stuff like minecraft servers; baremetal 12 node k8s cluster on opi5s for "research" (coz I do lots of k8s at work).
Each "stage" above is like incremental failure domains, unifi only keeps internet working, core vms add functionality (like unifi mgmt, rancher, etc), truenas is for "fun extras" etc. k8s lab has nothing I need to keep on it because distributed storage operators are still kind of explodey.
Like each part makes sense individually but when I look at the whole thing I start to question my mental health.
Imagine simplest possible deployment you've cooked up.
Now imagine explaining your mother how to maintain it after you're dead and she needs to access the files on the service you setup.
usually, selfhosting is not particularly hard. It's just conceptually way beyond what the average joe is able to do. (Not because they're not smart enough, but simply because they never learned to and will not learn now because they don't want to form that skill set. And I'm not hating on boomers, you can make the same argument with your hypothetical kids or spouse. The parents are just an easy placeholder because you're biologically required to have them, which isn't the case for any other familial relationship)
why does it have to be a non-technical next of kin ? Write down the details for a technically inclined person to follow, maybe a specific friend. Print at the top of the page “show this to X”. In the document explain how to recover the necessary data and replace the setup with a standard one.
I assume most people know at least one person who would do this for them , in the event of their death?
> I think about it in terms of what will happen when I die and what I will be inflicting on my family as the ridiculous setups stop working.
I run proxmox too and I've now got a nice little infra at home.
For my family it's simple: I explained them that the infra doesn't matter. The only thing that matter is data. And that there are many, many, many redundant backups of the data and that the backups are functional. That the data are correct (not a single bit missing) and pristine (deduplicated etc.).
Basically 20 years of family pictures and family movies, notarized documents, many proofs of big money transfer (as we now live in an hellhole of KYC/AML where I constantly need to prove money transfer, even when they're so old my banks don't allow me to get that info anymore), all the invoices related to real estate, medical stuff, cars, etc.
My backup procedure uses an intermediary steps that restore the data from the backup and verifies that the data is correct: once that step passes, the backup gets the greenlight. 3-2-1. Even more than 3-2-1.
Cryptographic hashes everywhere, including in filenames: I've got scripts that do verify x% of the files, random sampling style. I'm 100% guaranteed that at least 99.999% of the files are correct. And there are so many backups (online, offline, onsite, offsite, ...), all checksumed. My family won't lose our data. They're literally in various safe and in several countries.
Once I won't be there, they'll have the data up to that point. FWIW both my wife, daughter and brother --although they're not techies-- all happen to be familiar with computers and all took Python lessons. They know what a CLI is.
So even should they have a problem hooking up a hard disk, there's not a world in which they cannot do that:
"LLM> Dear AI, I've got disks with backups of family pictures and notarized documents, how can I access them?"
The world where you couldn't ask that question doesn't exist anymore.
Data of new memories shall be theirs to do deal with though.
P.S: besides that I think homelab'ing is also for convenience and understanding how things (like the network and servers) do work. It's to me more about thinkering and learning than "controlling".
I have been getting into this too. I caution anybody with self-hosting/tinkering tendencies against starting a tech company because it just makes it so much easier to justify this stuff...
Eventually serving a regular old container doesn't cut it anymore and you find yourself needing to pay these weird newspapers nobody reads to publish your business' alias because it's a requirement for a legal DBA which ASIN needs to let you get your own IPV6 block, which you need to truly own you and your customers' IPs and it's not worth becoming an AS without it, but then you can actually move towards physically become your own ISP and then...
The ingress problem people solve with tailscale is one of the hardest. I'm curious to see if it's possible to implement STUN/TURN [0-1] with a generally good mechanism for exposing the server to the Internet by caching all static files and blocking dynamic access to the backend with a loginwall, which authenticates allowed users with email "magic links" -> nonrenewable access tokens. In theory it should not be excessively difficult, expensive, or risky to do this.
It's just relevant enough to what we're doing with remote development environments for me to justify another rabbit hole
- one single machine
- nginx proxy
- many services on the same machine; some are internal, some are supposed to be public, are all accessible via the web!
- internal ones have a humongous large password for HTTP basic auth that I store in an external password manager (firefox built in one)
- public ones are either public or have google oauth
I coded all of them from scratch as that's the point of what I'm doing with homelabbing. You want images? browsers can read them. Videos? Browsers can play them.
The hard part is the backend for me. The frontend is very much "90s html".
Nice! I have a friend who is starting to program his infrastructure/services from scratch. It's a neat way to learn and make things fit well for your own needs
Impressive setup, but I would assume it to be very operations-intensive because of the high number of deployed components and their complex configuration. Plus, if you are serious about self-hosting, you would need the facilities and infrastructure to deploy it: server rack, redundant power supply, smoke detectors, fire extinguisher... I would never let my PC-grade hardware run unsupervised in my home.
And if I understood correctly, you would still have to have some server on the Internet for running your Headscale VPN, so you need your own dedicated Internet connection - ADSL, dial-up, cable modem would not be enough.
Great write-up. I have been tinkering with Immich over the last few months and go back and forth whether I want to just limit it to Tailscale for use away from home or if I want to go through the trouble of setting up a reverse proxy on a VPS. I think my biggest concern is finding a relatively user-friendly monitoring/security solution to alert me if anybody is trying some sort of attack against the VPS
I wish I had the time to do any of this. I could probably do it on a weekend but maintaining it, upgrading it to keep up with new releases would be something I wouldn't have time for.
I end up just paying a cloud provider and forget about it.
Anyone else on the same boat? What has been your approach?
Honestly, I self-host about a dozen services and upgrades take me less than a minute per month usually.
I simply have one folder per service, each folder contains a docker-compose stack and a storage directory. Updating is simply a matter of running `docker compose pull` and `docker compose up -d`. Nothing more.
Breaking updates requiring to tweak the config are very uncommon, and even when they happen it's only a few minutes of checking the updated config and applying it.
IMO this is the simplest way to self-host. No VM, no complex software install, nothing more than a simple Docker Compose setup that's fully automated.
That sounds similar to my setup, but each folder is a btrfs subvolume and my update script takes a snapshot before updating. I keep the Docker compose file together with the volumes in that subvolume.
If something breaks I can decide to figure out why, or revert.
With previous setups, I was certainly guilt of not upgrading and doing the maintenance needed. That's one reason why I like using NixOS and ZFS. Both provide really easy rollback options. So all I need to do is run an update and rebuild. If things work, no more for me to do. If things don't, I can try debugging or just revert back to the previous release till I have time to
But also I think using a cloud provider is fine if you're happy with the experience. It is a time sink to get things setup and it's not zero maintenance time. It's reasonable to weight those costs
It's usually not a single weekend. If you're like me, it starts out with thinking it'd be nice to install Plex on an old gaming PC. A year later, it has organically grown into a rube goldberg machine of proxmox and home automation. Which I guess just reinforces your point.
Joking aside, a minimal setup just using docker compose is pretty manageable. Self hosting many projects is as easy as 'docker compose up -d', and upgrades are straightforward as others have pointed out.
I like it. Why Flame, though? It’s built using node, react, redux… meaning you are bringing dozens (if not hundreds) of third party dependencies to your secure kingdom. I don’t think it’s worth it for the start page (could easily be a single html page with hardcoded links)
It's entirely because I've used it before. I just wanted something familiar to solve a problem quickly. I also think it looks nice. I'm not too worried about the security implications, since it is behind Tailscale and Authelia. I'm not committed to it, and do want to explore other options in the future
Why bother with SSO? If your family and closest friends use something like a wireguard client (iOS for example has a very good one that takes only a minute to configure permanently), the users simply switch a toggle and they are now on your private network and don’t need to SSO to anything (provided you have left everything open).
For a small home network the pros of that approach vastly exceed the cons.
The services we use, like Nextcloud or Mealie, are designed for folks to have their own user accounts. SSO means they can use the same login across all of them without me having to manage that for them (and also avoids me having to know their passwords). It does complicated the setup, but not the operation, and that makes it more likely folks will use the services
I self host about 20 separate apps. I’m in the middle of an SSO implementation project because I do not want to continue managing credentials for 20 separate apps.
I’ve considered opening some of these apps to family members, and having one place to deal with any auth issues is a high priority for me.
That’s ok. But step back further. Do you need to fine grain permission people on the majority of those apps? If you don’t, then SSO is more of a pain than it’s worth. Simply control the network access and leave the apps alone.
Nice writeup, thank you. I already thought about having NixOS on my server, but currently I prefer proxmox. There are projects with NixOS + Proxmox, but I did not test it yet.
> My main storage setup is pretty simple. It a ZFS pool with four 10TB hard drives in a RAIDZ2 data vdev with an additional 256GB SDD as a cache vdev. That means two hard drives can die without me loosing that data. That gives me ~19TB of usable storage, which I’m currently using less than 10% of. Leaving plenty of room to grow.
I would question this when buying a new system and not having a bunch of disks laying around... having a RAID-Z2 with four 10GB disks offers the same space as a RAID1 with two 20GB disks. Since you don't need the space NOW, you could even go RAID1 with two 10TB disks and grow it by replacing it with two 20TB as soon as you need more. This in my opinion would be more cost effective, since you only need to replace 2 disks instead of 4 to grow. This would take less time and since prices per TB are probably getting lower over time, it could also save you a ton of money. I would also say that the ability of losing 2 disks won't save you from having a backup somewhere...
Oh yeah, I don't think the way I went about it was necessarily the most cost effective. I bought half of them on sale one year, didn't get around to setting things up, then bought the other two a year later on another sale once I finally got my server put together. I got them before I had my current plan in place. At one point I thought about having more services in a Kubernets cluster or something, but dropped that idea
Also agree, RAID isn't a replacement for a backup. I have all my important data on my desktop and laptop with plans for a dedicated backup server in the future. RAID does give you more breathing room if things go wrong, and I decided that was worth it
I went through the same situation and noticed that modern hard drives are big enough for fitting RAID1 into nearly every homelab use case except high-res Video footage (including a bluray movie collection)
Two drives are easy to replace, easy to spare, consume less power and are quieter than 4+.
The only advantage i See in raid5/6 is on 25Tb of storage requirement within 3 years.
As OP says, I think this is the kind of thing that needs to be considered whenever the decision is made.
As another data point, my NAS runs 4x4TB drives. When I bought them new some 2-3 years ago, all at the same time, they were cheaper than buying the equivalent 2x8TB.
My situation was somewhat different, though, since I'm running raidz1. But I did consider running a mirror, specifically in order to ease upgrading the capacity. However, I didn't expect to fill them /that/ quickly and I was right: yesterday it was still less than 70% full.
You are right, but to be max cost effective you could have gone 2 * 4tb or 2 * 6tb for 18 months and then sell the drives still in warranty to upgrade to more storage...
Estimating storage growth is hard but when you monitor it regularly, its saving you much money
Maybe... When I was younger, I used to buy and sell computer stuff. Didn't have much money, so it kinda made sense. But it required me to keep up to date with prices, specs, figure what's the best value, follow markets and jump on an occasion, etc. It got old after a while. There's also value in getting something that just works for you, if it's not absurdly expensive, and forget about it. Do something else with my time.
I still love to tinker and set up a homelab and whatnot, but I don't care that much about hardware anymore. For my needs, if it's at least a 6th gen Intel and I can't hear it in my living room, it's good enough. The NAS lives in my parents' basement, so it can be somewhat louder (with 4 drives instead of two).
For this particular setup, my initial usage was above 4 TB, so I should have went with 2x6, which was /maybe/ cheaper (don't remember), but then it would have required me to deal with selling used gear and go through the motions of upgrading again. Doing this every 4-5 years? Sure. Every year? Hell no.
I've been tempted to use NixOS for my self hosted setup but I just can't bring myself to do it.
My setup is quite simple, it's just a few VMs with one docker compose file for each. I have an ansible playbook that copies the docker compose files across and that's it. There's really nothing more to it then that, and maintenance is just upgrading the OS (Fedora Server) once the version reaches EOL. I tend to stay 1 version behind the release cycle so upgrade whenever that gets bumped.
I do use nix-darwin on my macs so I do _see_ the value of using a nix configuration, but I find it difficult to see if the effort in porting my setup to Nix is worth it in the long run, configuration files don't get written in a short time. Maybe LLMs could speed this up, but I just don't have it in me right now to make that leap
> I've been tempted to use NixOS for my self hosted setup but I just can't bring myself to do it.
I recently tried nixos and after spending a week trying it out, I switched my home network and 2 production servers to nixos. It has been running as expected for 3,4 months now and I LOVE it. Migrating the servers was way easier than the workstations. My homeserver was setup in a few hours.
I also recently bought a jetson orin nano to play and learn on and I set up nixos with jetpack-nixos there too. I know with gentoo this would have been a (much) more painful process.
I have used gentoo for over 20 years and have always felt very much at home. What annoyed me was that the compile times on older computers were simply unbearable. Compiling GHC on my 2019 dell xps just takes 6 hours or something like that.
The big difference for me was NixOS provides really simple rollbacks if something goes wrong, where with Ansible and compose files, that's possible, but you have to do it yourself
But also if you're setup is working for you, I think that's great! It sounds like you have a good system in place
LDAP author here. I'm happy that LLDAP is mentioned and yet that it is not highlighted. The goal of the project was to have a simple LDAP server that is easy to install/manage for self-hosters without knowledge of LDAP required. Cheers and congrats on your setup!
> Authelia provides authentication and authorization for services in a couple of ways. For services that support OpenID Connect it provides a very simple single sign on experience. If not, Authelia can integrate with my reverse proxy (nginx) and require the user login before the reverse proxy allows access to a service.
Recently I found out Gitea or Forgejo can act as an Oauth provider. And since these support ldap you can for example deploy a Samba AD and set it up as an authentication source for Gitea/Forgejo. If you enable the OAuth feature you can connect stuff like grafana and log in with your Samba AD credentials.
To me this is more convenient than running a dedicated auth service considering Forgejo can also provide git, wiki, docker registry (also authenticated) and other function. It's such an underrated piece of software and uses so few resources.
Great blog post, but unfortunately from my experience with my kinda tech-friendly family, i can tell you that not exposing service publicly is an absolute UX killer.
Nobody uses the local nextcloud because they just don't think they can rely on it, it doesn't always work from their perspective, and is too finicky to use, because it needs an external app (Tailscale).
This can be only fixed when the app itself can trigger a vpn connection, and I don't think this is going to happen any time soon.
An easy solve for this is to buy public domains for the sites you want to use, run a static website on them that says “turn on tailscale to access this site, set that up here (link to download a client preconfigured for my tailnet, invite only of course)”, then use tailscale DNS overrides to set up CNAMEs for that public DNS’s (sub)domains which point to the tailnet internal service domains.
I do have to sit down and walk folks through setting up Tailscale, Nextcloud, etc on their devices. So far though, I haven't had any complaints once that is done. Nextcloud just syncs in the background and they can navigate to sites like normal. But my family is probably more tech literate than most, so that helps
This is kinda similar to something I'm trying to setup. I have most of my self-hosted infrastructure running in docker containers, but I want to put some stuff on a nixOS ec2 instance. Mostly services I want to never go down or be affected by my local network (uptime kuma) and chat stuff (irc bouncer, conduit, soju, etc etc).
I use nixOS on my laptop but don't make many nix projects, and TBH I have no idea how to test this setup locally before deploying it. I have some nix stuff setup that spins up a VM and exposes the ports on localhost, but it's brittle and rapidly spaghettifying. Do you have any tips for testing this stuff as part of a local project?
On my NixOS laptop I you can setup services I'm interested in trying, but just run them locally. So I don't setup things like SSL (you can, it sometimes just makes getting a new SSL cert for that same domain take some time). I just update my /etc/hosts to the local IP and can give that a go
For trying out the more complicated setup parts, like SSL, Tailscale, etc, I created a NixOS VM that I setup the same way I wanted for my "production" use case. Once I have the config file the way I wanted, it's as simple as moving it to my non test VM (baring previous mentioned SSL issues). And I only tested one part at a time, adding them together as I went
But also, one of the great things about NixOS is it's really easy to incrementally try things and rollback. Once I got the skeleton of the setup working, I've mostly done my testing on my "production" server without issue
I am curious what are some good enough cheapskate self-hosting setups?
I want to self-host one of those floss Pocket replacements but I don't want to pay more than what these projects charge for hosting the software themselves (~$5). I am also considering self-hosting n8n. I don't have any sophisticated requirements. If it was possible I would host it from my phone with a backup to Google Drive.
Any of the 1L PCs from Dell, HP, or Lenovo. They sip power (5~10 watts), and take up minimal space. I've got a 6 or 7 VMs running on one, and it barely breaks 5% CPU usage.
This, when I was a student and had to live frugal (2001-2008 or so), I got a second-hand Dell, put it on top of a high cupboard in my dorm room, and installed a bunch of services (e.g. Trac was very popular in the day for hosting projects).
It won't give you 99.999% uptime, but for that stage in my life it was just stellar. I even had an open source project (Slackware fork) where I collaborated with someone else through that little machine.
Second-hand hardware is also a great way to get high-quality enterprise hardware. E.g. during the same time period I had a Dell workstation with two Xeon CPUs (not multi-core, my first SMP machine) and Rambus DRAM (very expensive, but the seller maxed it out).
Agree. If low cost and maximum value is you're goal, grab a used one of these or similar speed laptop (and you sort of get battery back up in that case)
Really, any machine from the last decade will be enough, so if you or someone you know have something lying around, go use that
The two main points to keep in mind are power draw (older things are usually going to be worse here) and storage expandability options (you may not need much storage for your use case though). Worse case you can plug in a USB external drive, but bare in mind that USB connection might be a little flaky
Me too. It works really well now that I have a silent version with SSD and passive heat sink. This avoids my son turning it off because the fan noise annoys him. I am thinking about adding kubernetes for failure resilience but that is a work in progress.
I'm not using the router, but I do like reading about Stapelberg's quest for ridiculous home Internet speeds! I make do with standard gigabit fiber haha
Curious what the driver for nixos and packages over docker was. Docker was the huge step up for me in making the homelab easy to run, update, and recover when things failed. It also made managing services endpoints and ports remote easier than when they all lived on the operating system. Wish this was delved into a little more in the post.
I can touch on it more. Docker and compose files are great for getting things going, contained, and keeping everything declarative
But I found the more services I used with Docker, the more time it took to update. I didn't want to just update to latest, I wanted to update to specific version, for better rollback. That meant manually checking and updating every single service, bringing each file down, and then back up. It's not entirely unmanageable, but it became enough friction I wasn't updating things consistently. And yes, I could have automated some of that, but never got around to it
NixOS, in addition to the things I mention in the post, is just a two step process to update everything (`nix flake update` and `nixos-rebuild`). That makes updating my OS and every package/service super easy. And provides built in rollback if it fails. Plus I can configure things like my firewall and other security things in NixOS with the same config I do everything else
Also, Nix packages/services provides a lot of the "containerization" benefits. It's reproducible. It doesn't have dependency problems (see this for morehttps://nixos.org/guides/how-nix-works/). And most services use separate users with distinct permissions, giving pretty good security.
It's not that Docker can't do those things. It's that Nix does those things in a way that work really well with how I think
A pain point you mention is that everyone must run the tailscale client. Have you considered exposing everything on the public internet using something like Cloudflare Tunnels? You can have cloudflare handle auth on their edge network, which mitigates the worry about having to deal with 0-days on a self hosted auth solution. You have a pretty sophisticated directory setup tho so I'm not sure how well this would fit in with the existing infra.
It is something I considered. Ultimately I didn't want to depend on Clouflare (or any other provider) for something as core to my setup as my remote access
But it's a totally valid option, just not one that fit with my preferences
Is there home lab for isolated LAN and "self-sufficient" devices?
I want to have a block of gunk on the LAN, and to connect devices to the LAN and be able to seamlessly copy that block to them.
Bonus: any gunk I bring home gets added to the block.
First part works with navidrome: I just connect through the LAN to my phone with amperfy and check the box to cache the songs. Now my song gunk is sync'd to the phone before I leave home.
This obviously would fit a different mindset. Author has a setup optimized for maximum conceivable gunk, whereas mine would need to be limited to the maximum gunk you'd want to have on the smallest device. (But I do like that constraint.)
I woke up today with a plan of making my DNS at a separate site work properly with ipv6, over my wireguard. I use ULAs for the point to point wireguard link, and GUAs don't like routing to ULAs. I figured the choice was between routing my two sites GUAs over the wireguard when talking to each other, or deploy ULAs in my networks. 4hrs later I had everything set up with ULAs. Had lunch. Decided that was awful. 3hrs after that I've got my GUAs going over the wireguard.
In case the author is around: On mobile (Chrome on Android) the screenshot is not readable at all and there is also no way to open an enlarged version, let alone zoom into the page.
I too use LLDAP and Authelia. I use Caddy (no Traefik) as a reverse proxy to protect my services using 2FA SSO. It's very easy to use and I can access all my services anywhere in the world without bothering with a VPN.
It's not open-source or self-hosted but putting it out there: CloudFlare zero-trust is amazing and free. In my setup, I have a cloudflared tunnel configured in my homelab machine and I expose individual services without a VPN or opening up my firewall. You can also set up authentication with SSO, and it happens before reaching the backend application which makes it more secure. This is easy for family and friends to use, because they don't need to setup anything from their side, just go to the URL and login.
https://developers.cloudflare.com/cloudflare-one/connections...
I seriously don’t understand why would people choose this over not exposing anything at all, except for Wireguard port. I have my client to automatically connect my home LAN when I’m not on WiFi and get access to all my self-hosted services without risking anything. You rely on third party solution which may or may not be made to government agencies. You also need to trust they Cloudflare doesn’t make mistakes, either.
Also, how do you configure Cloudflare for a road warrior setup? How do you track ever changing dynamic IPs? As mentioned, all I need is a Wireguard client and I’m golden.
CloudFlare zero-trust is very good, but i thought you need to have Cloudflare as man-in-the-middle on your domain to have this authentication flow work? ie. the TLS certs needs to live with Cloudflare.
Yeah, that is how I use it. You can technically host any TCP including end to end encrypted data through CloudFlare tunnels but you need the cloudflared app installed on the client side to access it (SSO still works even for this case). I find having to manage certificates and installing cloudflared everywhere is too much of a hassle.
I understand that proxing through CloudFlare gives them a lot of visibility and control, but I find that risk acceptable given my application.
Setup your own WireGuard rather than Tailscale.. this is too much like Authy delegating AAA to a third-party.
- Store your SSH public keys and host keys in LDAP.
- Use real Solaris ZFS that works well or stick with mdraid10+XFS, and/or use Ceph. ZoL bit me by creating unmountable volumes and offering zero support when their stuff borked.
- Application-notified, quiesced backups to some other nearline box.
- Do not give all things internet access.
- Have a pair (or a few) bastion jumpboxes, preferably one of the BSDs like OpenBSD. WG and SSH+Yubikey as the only ways inside, both protected by SPA port knocking.
- Divy up hardware with a type 1 hypervisor and run kubernetes inside guests in those.
- Standardize as much as possible.
- Use configuration and infrastructure management tools checked into git. If it ain't automated, it's just a big ball of mud no one know how to recreate.
- Have extra infrastructure capacity for testing and failure hot replacements.
Annoying thing about WireGuard is their outdated and buggy iOS client. When you set up a dns with A and AAAA it'll prefer the A address, even when you're on a 646xlat network, so now that connection is proxied and will time out after a while.
Yep. Other reasons I had to go for IPv4 only a while despite everything else being dual stack. "Argh!" at that one vendor who can't get their act together.
How can one run vanilla wireguard and leverage features offered by headscale? At minimum, a bunch of bash scripts would do the exact same thing, if not worse
Don't do it with bash. You can at least use Ruby, Python to make an API for it, or use configuration management. They really didn't think about being (local) runtime configurable for the dev/ops UX being too ultra *NIX purist with single file plain text configuration. At least it could have a plain text watch directory like daemontools for dynamic reconfiguration.
im using proxmox but struggling to setup subnets and vms
should I be using terraform and ansible?
im using cursor to ssh and it constantly needs to run commands to get "state" of the setup.
basically im trying to do what I used to do on AWS: setup VMs on private network talking to each other with one gateway dedicated to internet connection but this is proving to be extremely difficult with the bash scripts generated by cursor
if anyone can help me continue my journey with self hosting instead of relying on AWS that would be great
> im using proxmox but struggling to setup subnets and vms
That is a pretty broad target. I would say start by setting up an opnsense vm, from there you can do very little to start, just lock down your network so you can work in peace. But it can control your subnet traffic, host your tailscale, dchp server, and adguard home, etc.
As somebody who was quite used to hosting my own servers, before I first set up my homelab I thought proxmox would be the heart of it. Actually opnsense is the heart of the network, proxmox is much more in the background.
I think proxmox + opnsense is great tech and you should not be adding in terraform and ansible, but I am not sure that using cursor is helping you. You need a really good grasp of what is going on if your entire digital life is going to be controlled centrally. I would lean heavily on the proxmox tutorials and forums, and even more on the opnsense tutorials and forums. Using cursor for less important things afterwards, or to clarify a fine point every once in a while would make more sense.
I've found a lot of docs (Proxmox and TrueNAS are both guilty of this) assume you have existing domain or tool knowledge. I'd recommend checking out some videos from selfhosting YouTubers. They often explain more about what's actually happening than just what buttons to select
Also, I found TrueNAS's interface a little more understandable. If Proxmox isn't jiving with you, you could give that a try
I agree Proxmox default networking is lacking/insufficient at best. If you have VLANs or want to do LACP, anything more advanced than a simple interface you'll run into the limitations of the Proxmox implementation quite quickly.
I think the networking experience for hosts is one of the worst things about Proxmox.
Handle subnets on your router, then in ProxMox make the primary network interface you'll be passing to VM's or Containers VLAN aware, with the VLAN tags that it'll support defined and you're good to go.
Try using Proxmox's web UI to create a Linux Bridge for each subnet, then attach VMs to appropriate bridges and configure a VM with two interfaces as your router between networks.
As for Nextcloud vs a restic server, Nextcloud is heavier, but I do benefit from it's extra features (like Calendar and Contact management) as well as use a couple of apps (Memories for photos is quite nice). Plus it's much more family friendly, which was a core requirement for my setup
As you said, everybody will have their own approach. For me, setting up a dedicated firewall to protect the soft underbelly of my less tech savvy family computers was the primary motivation for starting a home lab. I could not imagine a network architecture without it.
I found the Tailscale client experience is quite nice and headscale had built in OIDC support (so easy auth for my users)
If I started this setup later I might have also used pangolin, which also provides a nice management interface on top of WireGuard https://github.com/fosrl/pangolin
It’s a shame he doesn’t self host an internal mail server at least with restricted outbound/smtp.
Something like this is very easy to setup with projects such as stalwart which also offers CardDAV and CardDAV (think easy synchronization of calendar and contacts without relying on "cloud").
He already has tailscale + headscale, adding in an internal only mail/collaboration server would be a win.
Hey, I ruled out a mail server for external, since I've heard many people have issues with other providers (Gmail, Outlook, etc) randomly blocking email. Didn't feel I could rely on it
Having an internal only mail server for notifications is an interesting idea. I've been using ntfy and Matrix to achieve something like that, but not all services support those notification methods. I'll keep that in mind!
It's nice to see a home lab on HN. Hardware has become a lost art for many.
If you dont have a home lab, start one. Grab a 1l pc off of ebay. Think center m720q or m920q with an i5 is a great place to start. It will cost you less than 200 bucks and if you want to turn it into a NAS or an Opnsense box later you can.
When it arrives toss Proxmox on it and get your toys from the community scripts section... it will let you get set up on 'easy mode'. Fair warning, having a home lab is an addiction, and will change how you look at development if you get into it deeply.
I credit homelabbing through my twenties with just about everything good that's happened to me in my career. I certainly didn't end up being moderately employable because I'm smart, charismatic, incisive, creative, lucky, educated, diligent, connected, handsome, sanitary, interesting, or thoughtful; no, it's because I have a tendency toward obsession, delusions of grandeur, and absolutely terrible impulse control.
So I started buying junk on eBay and trying to connect it together and make it do things, and the more frustrated I got, the less able I was to think about literally anything else, and I'd spend all night poking around on Sourceforge or random phpBBs trying to get the damn things to compile or communicate or tftp boot or whatever I wanted them to do.
The only problem was eventually I got good enough that I actually _could_ keep the thing running and my wife and kid and I started putting good stuff on my computers, like movies and TV shows and music and pictures and it started to actually be a big deal when I blew something up. Like, it wasn't just that I felt like a failure, but that I felt like a failure AND my kid couldn't watch Avatar and that's literally all he wanted to watch.
So now I have two homelabs, one that keeps my family happy and one that's basically the Cato to my Clouseau, a sort of infrastructural nemesis that will just actually try to kill me. Y'know, for fulfillment.
Not sure if it happens to most, but I have looped back around to not wanting to play sysadmin at home. Most of the stuff I have running I haven't updated in a awhile, luckily since I own it and it's all internal I don't need to worry about anyone taking away my locally hosted apps. Thank the IT gods for docker compose, and tools like portainer to minimize the amount of fuddling around I have to do.
That, and I've learned the stuff I need to from my homelab. Earlier in my career, setting up a light version of a production network at home was hugely educational and I think it was a large part of why I'm a senior today. But now, I don't need to run all that at home unless I have specific learning objectives. So I keep my home network a lot simpler than it used to be, as a result.
Same, replaced the ISP router with my own and have a single box which has storage and compute for running VMs and NFS and that is it. Last thing I want to be doing on a Friday night is debugging why my home network is broken.
Why people need these overly complicated setups and why do they need to have an access point to reach their "den" from anywhere is beyond me. People and their digital gadget delusion.
Security paranoia, but here are the details of my home lab. WHY? If god forbid someone gets in they could in an instant identify the target...
I think that's a fair point. Kinda like with Arch, you do have to know what you want to setup NixOS right now
I really like what's happening in the ublue space were folks are tweaking and optimizing distros for specific use cases (like bazzite for gaming) and then sharing them
NixOS does support that to an extent, but it certainly doesn't have the same community movement behind it like those
IMO this is too complicated. I think products like the Synology Disk Station strike a better balance between ownership of data and maintenance over time. Tailscale even publishes a client for Synology products.
Everyone will have different goals and preferences. For instance, my dad just wanted a way to backup and remotely access some files, so we got him a Synology NAS. It's great for it's target users and if you're one of them, awesome!
I just don't like the lock-in that you get Synology. Plus I do enjoy tinkering with these things, so I wanted to put together something that balances usability, complexity while minimizing that lock-in
Is this the same Synology that a few years ago numerous people had their boxes ransomwared when they were open to the public internet. Synology continues to be shit value for the tools you get and as much as I want the convenience of ready to go software and hardware they cannot be relied on.
I am thinking about replacing my Synology BUT I have had three and now ECC over the last 20 years and they have done their job faullessly. There are stupid things now like complaining about non Synology ram modules.
> Relatively easy for family and friends to use
> This means keep one login per person, ideally with SSO, for as many services as I can
Truly S-tier target. Incredible hard, incredible awesome.
I've said for a long time that Linux & open source is kind of a paradox. It goes everywhere, it speaks every protocol. But as a client, as an end. The whole task of coordinating, of groupwareing, of bringing together networks: that's all much harder, much more to be defined.
Making the many systems work together, having directory infrastructure: that stuff is amazing. For years I assumed that someday I'd be running FreeIPA or some Windows compatible directory service, but it sort of feels like maybe some OpenID type world might possibly be gel'ing into place.
Appreciate that! Simple login and access was certainly the hardest requirement to hit, but it can be the difference between people using something and not
And I agree with the feeling that open source is everywhere, up until a regular user picks up something. I think part of the paradox you mention is that every project is trying to work on their own thing, which is great, but also means there isn't a single entity pushing it all in one direction
But that doesn't mean we can't get to nice user experiences. Just in the self-hosting space, things have gotten way more usable in the last 5 years, both from a setup and usage perspective
I completely agree with the paradox, just yesterday I posted how FOSS is not accessible to non-techies on my problem validation platform[1].
I've been thinking if a platform which connects techies to non-techies can help solve that, say like a systems integrator for individuals.
[1] https://needgap.com/problems/484-foss-are-not-accessible-to-...
It's not supposed to be. You put in time, use your brain to understand the system. Even a non-techie can easily understand OIDC and Oauth2, it's not that hard
As a techie, experienced in security, reading the OIDC spec... there are definitely some things I don't understand in there. I'm not sure the authors even understand what's going on.
On 2023-12-15 they published an update to OpenID Connect Core 1.0, called "errata set 2". Previously it said to verify an ID token in a token response, the client needs to
> * If the ID Token contains multiple audiences, the Client SHOULD verify that an azp Claim is present.
> * If an azp (authorized party) Claim is present, the Client SHOULD verify that its client_id is the Claim Value.
The new version is quite different. Now it says
> * If the implementation is using extensions (which are beyond the scope of this specification) that result in the azp (authorized party) Claim being present, it SHOULD validate the azp value as specified by those extensions.
> * This validation MAY include that when an azp (authorized party) Claim is present, the Client SHOULD verify that its client_id is the Claim Value.
So core parts of the security of the ID Token are being changed in errata updates. What was the old purpose of azp? What is the new purpose of azp? Hard to tell. Did all the OIDC implementations in existence change to follow the new errata update (which didn't update the version number)? I doubt it.
https://openid.net/specs/openid-connect-core-1_0.html
https://web.archive.org/web/20231214085702/https://openid.ne...
Or how about a more fundamental question: Why does the ID Token have a signature? What attack does that signature prevent? What use cases does the signature allow? The spec doesn't explain that.
> Why does the ID Token have a signature?
The ID Token can be passed from the Identity Provider to the Relying Party (RP) in a few ways.
When `response_mode=id_token` is used, the ID Token can be passed in the front channel directly to the RP during a browser redirect. Since the ID Token is coming from the browser, it must be signed to ensure that a malicious actor can't tamper with it. Otherwise an actor could swap out a `sub` or an `email` claim and the RP would be none the wiser.
The ID Token can also be returned from the `/token` endpoint after exchanging an authorization code. Since the `/token` endpoint is a back channel call over HTTPS, the ID Token doesn't necessarily need to be signed here to avoid tampering. The RP can trust that TLS gets the job done. However, there are substantial benefits to having it be signed:
- If ID tokens were only signed sometimes, we'd have two different standards for how to construct and handle an ID Token, which would be quite confusing.
- Signed ID Tokens can be passed around to assert identity to other entities within the system. For example, there are some promising draft specifications that explore exchanging ID Tokens for access tokens in other systems. This is only possible because the ID Token cannot be tampered with.
https://datatracker.ietf.org/doc/draft-parecki-oauth-identit...
>Since the ID Token is coming from the browser, it must be signed to ensure that a malicious actor can't tamper with it. Otherwise an actor could swap out a `sub` or an `email` claim and the RP would be none the wiser.
Are you referring to a login CSRF attack? Where an attacker causes a victim to visit a URL that forces the victim to log in to an account of the attacker's choosing? A signature on the token doesn't solve that AFAICT. The reason a token signature doesn't solve that is that an attacker can take the attacker's own ID Token that has a valid signature, and do the attack with that ID Token, forcing the victim to log in to the attacker's account.
Yes, the token signature does reduce the attacker's ability. Without a token signature, the attacker could force the victim to log in to any account. With a token signature the attacker can only force the victim to log in to the attacker's account. But this isn't a full solution. Some other anti-CSRF mechanism needs to exist. Once that anti-CSRF mechanism exists, the token signature is no longer useful.
Adding a signature to the token doesn't make sense to me as a login CSRF protection mechanism. If the designer's goal was to protect against login CSRF, the signature should be over something else, not just the token, and not part of the token. Then the token itself doesn't need the complexity of a signature (and the authorization code flow doesn't need the complexity of a signature), just the redirect flow needs it. Or even simpler, the designers could have mandated the authorization code flow, and not need any signature at all.
Or maybe you're referring to a completely different attack, where there isn't a victim user, just an attacker wanting to log in as someone else in the attacker's own browser. In that case, the signature doesn't solve that attack, because since everything happens in the attacker's browser, the attacker can modify the js locally to disable signature verification.
>- Signed ID Tokens can be passed around to assert identity to other entities within the system. For example, there are some promising draft specifications that explore exchanging ID Tokens for access tokens in other systems. This is only possible because the ID Token cannot be tampered with.
Yes, but the spec doesn't mention this. Just from reading the spec, a reader can wonder, "why is there all this unexplained complexity for no apparent reason?"
> Did all the OIDC implementations in existence change to follow the new errata update (which didn't update the version number)?
I mean, both the old and new version (at least, the parts quoted upthread) are exclusively SHOULD and MAY with no MUST, so (assuming, for the SHOULDs, the implementer had what they felt was sufficiently good reason) literally any behavior is possible while following the spec.
I think I could handle up to 20 users with `.htaccess` and just handing out passwords to my friends, actually
This seems like a very naive perspective.
It's not really that hard to be honest. If you are not dead set on specific services, but make sso compatability the main selection metric for the services it's very feasible and not that difficult. I had very little experience when I set up my self hosted system and was set up very quickly using caddy and authentik. Alternatively yunohost is a very easy to use distribution that sets up everything using SSO.
Agreed. I use Cloudflare Zero Trust for SSO authentication (I use Github for my self and Google/Apple for family) and Cloudflare Tunnels to expose homelab services behind NAT. It took an afternoon to set up the first service and adding services via IaC like terraform is easy.
The only time consuming thing since then has been figuring out how to use the Cloudflare auth header to authenticate with each individual app (and many don’t have multiuser capability so it’s not a big deal).
Yuno-host doing SSO is something I deeply deeply respect & love about it! Yes!
I haven't pulled it down yet to find out just how extensive the SSO is. My understanding is that it's not totally universal... But that it is like ~50%, which is super not bad given how many packages it has!
I use authentik with sso via Google, discord or GitHub.
It's good enough for everyone.
Hey y'all, I know getting a setup that feels "right" can be a process. We all have different goals, tech preferences, etc.
I wanted to a share my blog post walking through how I finally built a setup that I can just be happy with and use. It goes over my goals, requirements, tech choices, layout, and some specific problems I've resolved.
Where I've landed of course isn't where everyone else will, but I hope it can serve as a good reference. I’ve really benefited from the content and software folks have freely shared, and hope I can continue that and help others.
How are you finding Nix for the homelab to be? Every time I try it I just end up confused, maybe next time will be the charm.
The reason I ask is I homelab “hardcore”; i.e. I have a 25U rack and I run a small Kubernetes cluster and ceph via Talos Linux.
Due to various reasons, including me running k8s in the lab for about 7 years now, I’ve been itching to change and consolidate and simplify, and every time i think about my requirements I somehow end up where you did: Nix and ZFS.
All those services and problems are very very familiar to me, feel free to ask me questions back btw.
I certainly didn't take to Nix the first few times I looked at it. The language itself is unusual and the error messages leave much to be desired. And the split around Flakes just complicates things further (though I do recommend using them, once you set it up, it's simple and the added reproducibility gives nice peace of mind)
But once I fully understood how it's features really make it easy for you to recover from mistakes and how useful the package options available from nixpkgs are, I decided it was time to sink in and figure it out. Looking at other folks nix config on GitHub (especially for specific services you're wanting to use) is incredibly helpful (mine is also linked in the post)
I certainly don't consider myself to be a nix expert, but the nice thing is you can do most things by using other examples and modifying them till you feel good about it. Then overtime you just get more familiar with and just grow your skill
Oh man, having a 25U rack sounds really fun. I have a moderate size cabinet I keep my server, desktop, a UPS, 10Gig switch, and my little fanless Home Assistant box. What's yours look like?
I should add it to the article, but one of my anti-requirements was anything in the realm of high availability. It's neat tech to play with, but I can deal with downtime for most things if the trade off is everything being much simpler. I've played a little bit with Kubernetes at work, but that is a whole ecosystem I've yet to tackle
>The language itself is unusual and the error messages leave much to be desired. And the split around Flakes just complicates things further
Those are my chief complaints as well, actually. I never quite got to the point where I grasped how all the bits fit together. I understand the DSL (though the errors are cryptic as you said) and the flakes seemed recommended by everyone yet felt like an addon that was forgotten about (you needed to turn them on through some experimental flag IIRC?).
I'll give it another shot some day, maybe it'll finally make sense.
>Oh man, having a 25U rack sounds really fun. I have a moderate size cabinet I keep my server, desktop, a UPS, 10Gig switch, and my little fanless Home Assistant box. What's yours look like?
* 2 UPSes (one for networking one for compute + storage)
* a JBOD with about 400TB raw in ZFS RAID10
* a little intertech case with a supermicro board running TrueNAS (that connects to the JBOD)
* 3 to 6 NUCs depending on the usage, all running Talos, rook-ceph cluster on the NVMEs, all NUCs have a Sonnet Solo 10G Thunderbolt NIC
* 10 Gig unifi networking and a UDM Pro
* misc other stuff like a zima blade, a pikvm, shelves, fans, ISP modem, etc
I'm not necessarily thinking about downsizing but the NUCs have been acting up and I've gotten tired of replacing them or their drives so I thought I'd maybe build a new machine to rule them all in terms of compute and if I only want one host then k8s starts making less sense. Mini PCs are fine if you don't push them to the brim like I do.
I'm a professional k8s engineer I guess, so on the software side most of this comes naturally at this point.
400 TB?! do you collect Linux ISOs or are you doing photography?
Linux ISOs and backups.
> Oh man, having a 25U rack sounds really fun.
For certain definitions of the word “fun,” yes. I have a 35U (I don’t need that many slots, but at the time I did need it tall enough that my kids couldn’t reach the top, where I put the keys), with:
* 3x Dell R620
* 2x Supermicro (one X9, one X11)
* 1x APC UPS w/ external battery
* Unifi UDM Pro
* Unifi Enterprise 24-port switch
The Dells have Samsung PM863 NVMe drives which are used by Ceph (managed by Proxmox), with traffic sent over an Infiniband mesh network via Mellanox ConnectX3-Pro.
The Dells run K3OS in a VM, which is a dead project. Big mistake there.
The Supermicros have various spinners, and are in a ZFS pool. One of them is technically a backup that should power up daily to ingest snapshots, then power off, but there’s been some issue preventing that, so…
It was all very fun to set up, and has been eminently reliable, but it’s a bit much. While you can in fact make R620s relatively quiet, they’re still 1U, and those little 40mm fans are gonna whine. It’s background noise to me, but guests definitely mention it if we’re close to the rack.
Also, I’m now in the uncomfortable position of being stuck on Proxmox 7, because v8 (or more accurately, the underlying Debian release) dropped support for my HBAs, so the NAS would be dead in the water. I mean, I could compile my own kernel, or perhaps leverage DKMS, but that somewhat defeats the purpose of having a nice AIO like Proxmox. Similarly, my choose of K3OS means at some point I need to spend the time to rip everything out and start over with Talos.
Or - just maybe - I’ve done enough playing, and I should simply buy a JBOD chassis and a relatively new and quiet server (4U under light load means you can get away with much quieter fans), and just run stuff in Docker or gasp systemd. Or, hell, single-node K8s. My point is that it is fun, but eventually your day job becomes exhausting and you tire of troubleshooting all the various jank accumulating at home, and you stop caring about most of it.
I turned my supermicro 847 X10DRI+ into a JBOD by ripping out the mobo, the fans, and building a 140mm fan wall (zip ties and later 3D printed). I can highly recommend the move, it's dead silent and I have drive temps between 33 and 45C with 20 to 25 drives (I honestly don't know the total at this point).
As for your OS issues I also used to run Proxmox with Ceph (btw did you know you can use Proxmox's ceph with rook-ceph so you don't need 2 layers of storage?) including for the router and NAS, but I gave it up due to unwarranted complexity and went bare metal.
Don't know what would fit your particular use case but I can say this: I'm very happy I made a separate box with a supermicro X11 and a JBOD besides it, I can recommend this too; what benefit is there really to virualizing a NAS?
Regarding K3OS you're in luck - kubernetes manifests (you've got GitOps, right?!) are so portable you can just rebuild on a new OS. Give Talos Linux a spin. Again really think about why is it that you're virtualizing here too; maybe you genuinely need it, maybe not.
> Ceph / Rook
Not sure I follow here; why would I want Rook involved? I generally don’t want my orchestration layer - which is also consuming storage - to be involved with the management of said storage.
> Virtualizing vs. bare-metal
It’s partially to make upgrades via new base images easier. I bake images with Packer + Ansible, and so can push a new one out quite easily. The other part is that my NAS consumes very little compute resources, since it isn’t hosting any apps, so it would be a waste to run it on bare metal. Tbf I could run everything consuming the disk storage directly on it, but I had shied away from that initially and it stuck.
> GitOps
I have Helm templates, no ArgoCD. It’s been a TODO for me for quite some time. Same with Talos (I actually do have it running in parallel right now, just not hosting anything). My issue is that I am obsessed with getting things perfect, and for me that means bootstrapping from a bare VM to ArgoCD ingesting manifests and spinning up pods, all automated. I know this is possible, as I’ve seen it done, but I rarely have the time or energy to pursue it after work. I should probably get over myself and just manually install stuff so it’s functional.
I've been trying to switch my home cluster from Debian + K3s to Talos but keep running into issues.
What does your persistent storage layer look like on Talos? How have you found it's hardware stability over the long term?
>What does your persistent storage layer look like on Talos?
Well, for its own storage: it's an immutable OS that you can configure via a single YAML file, it automatically provisions appropriate partitions for you, or you can even install the ZFS extension and have it use ZFS (no zfs on root though).
For application/data storage there's a myriad of options to choose from[0]; after going back and forth a few times years ago with Longhorn and other solutions, I ended up at rook-ceph for PVCs and I've been using it for many years without any issues. If you don't have 10gig networking you can even do iSCSI from another host (or nvmeof via democratic-csi but that's quite esoteric).
>How have you found it's hardware stability over the long term?
It's Linux so pretty good! No complaints and everything just works. If something is down it's always me misconfiguring or a hardware failure.
[0] https://www.talos.dev/v1.11/kubernetes-guides/configuration/...
> after going back and forth a few times years ago with Longhorn and other solutions, I ended up at rook-ceph for PVCs
Curious to know what issues you ran into with Longhorn.
It was years ago but I recall high CPU usage being an issue in particular.
In general it's just not as battle tested as ceph and I needed something more bulletproof.
However I will say this: I'm sure that issue with the CPU usage was fixed (I was watching the GitHub issue) and you might not need your distributed FS to be CERN ready for your lab; AND the UI and built-in backups Longhorn offers are great for beginners so I'd suggest giving it a try if you don't already know you want ceph or OpenEBS Mayastor for the performance and so on.
Talos is the Linux kernel at heart, so.. just fine.
Honestly, I personally like the combination of keepalived+docker(swarm if needed)+rsync for syncing config files. keepalived uses VRRP, which creates a floating IP. It's extremely lightweight and works like a charm. You won't even notice the downtime, the switch to another server IP is instant
Keepalived is great. Learning about it was one of the best things I got from building HA-aiming infra at a job once.
Did you come across or consider using coolify at any point? I've been using it for over a year and quite enjoyed it for it's Heroku type ease of use and auto deployments from GitHub.
https://coolify.io/
No I haven't heard of it before. I do like the idea though, especially for side projects. Thanks for sharing, I'll look more at it!
Check out Dokploy too! https://dokploy.com/
Their license is still ambiguous, and I don't like how they communicate with those who inquire about it.
Do you use encrypted ZFS?
I have dabbled before with FreeIPA and other VMs on a Debian host with ZFS. For simplicity, I switched to running Seafile with encrypted libraries on a VPS and back that up to a local server via ZFS send/receive. That local server switches itself on every night, updates, syncs and then goes into sleep again. For additional resiliency, I'm thinking of switching to ZFS on Linux desktop (currently Fedora), fully encrypted except for Steam. Then sync that every hour or so to another drive in the same machine, and sync less frequently to a local server. Since the dataset is already encrypted, I can either sync to an external drive or some cloud service. Another reason to do it like this is that storing a full photo archive within Seafile on a VPS is too costly.
Yes! On top of the data safety features of ZFS, the fact you can encrypt a dataset and incremental send/receive is a fantastic ability
I appreciated the in depth look and while some ideas from your setup will take more time to implement, I just added flame for the dashboard and see how it fares with family.
Thank you! It's all a journey, hope flame works well for you!
Hi! Really excited by your work! I'm working on a similar project built on NixOS and curious what you thing.
My goal is to have a small nearly zero-conf apple-device-like box that anyone can install by just plugging it into their modem then going through a web-based installation. It's still very nascent but I'm already running it at home. It is a hybrid router (think OPNSense/PFSense) + app server (nextcloud, synology, yunohost etc). All config is handled through a single Nix module. It automatically configures dynamic DNS, Letsencrypt TLS certs, and subdomains for each app. It's got built in ad blocking and headscale.
I'm working on SSO at the moment. I'll take a look at your work and maybe steal some ideas.
The project is currently self-hosted in my closet:
https://homefree.host
Oh that sounds really rad! Certainly could have it's use cases. I really appreciate how NixOS enables projects like this. Best of luck with it!
Sometimes when I think about my home network, I think about it in terms of what will happen when I die and what I will be inflicting on my family as the ridiculous setups stop working. Or like, how much it would cost a police forensics team to try to make any sense of it.
I think "home labbing" fulfils much the same urge / need as the old guys (I hate to say it but very much mostly guys) met by creating hugely detailed scale model railways in their basement. I don't mean that in a particularly derogatory way, I just think some people have a deep need for pocket worlds they can control absolutely.
I ahve exactly the same thoughts and I wrote a document to be used in case I die.
Part one is money and where the important papers are.
Part twonis hiw to dulb down my home. How to remove the smart switches (how to wire back the traditionnal switches). How to mive self hosted key services to the cloud (bitwarden, mostly) and what to pay for (domain and mail). Hiw to remove the access point and go back to the isp box.
My wife is not supportive of the smart stuff but now that she knows she can dumb it down she is fine. Honestly she does not realize what strp back the lack of all this stuff will be. But at least it won't be my problem anymore:)
I have our family pictures on a RAID 1 array in my home lab. Every night they are rsynced to an external drive on a little computer at my in-laws. Both as a backup, and as an "if something happens to me" easy access. My wife doesn't have any interest in tech, so I wanted to make accessing it "just in case" as straightforward as possible. I told her that that is where all the photos are, and that it's just a USB drive she can connect to her laptop in case something happens.
Just ignore the useless threat scenario of someone stealing your physical disks to obtain your precious family data and you will be fine. In other words, just store all the photos and important documents in clear, some written down instructions and you should good to go.
I'm more worried by home automation in my case ^^;
The chance of someone breaking in to steal your sensitive files is next to nil I agree.
The chance of someone breaking into your house is sadly much more likely, and them choosing to take any computers they see is almost a certainty at that point.
Your drives are unencrypted. What's your next step if you come home tonight and find the house ransacked and the server gone?
It can be a side-effect. Francis Ford Coppola lost all his family photos when his PC was stolen in a burglary while in Argentina to shoot “Tetro”. Of course, he didn’t have backups.
> Your drives are unencrypted. What's your next step if you come home tonight and find the house ransacked and the server gone?
My drives are encrypted and so are my backups (with backups everywhere). But they're symmetrically encrypted with a password. The backup procedure contains a step verifying that decryption works.
Family knows the password: password is stored at different places on laminated paper (friends and family) but not alongside the backups.
Decryption of the backups is one command at the CLI (both brother and wife knows how to use a CLI and soon the kid shall too: already dabbled with it).
The one command is explained alongside the password, on the same laminated paper as the backups.
Yup I did really think this out, including rehearsals where I, literally, fake my own death (I fake a heart attack) in front of my brother and wife and I have to shut the fuck up while they open a CLI, hook up one of the backup hard disk and decrypt the backups.
Once a year we rehearse.
That way they are confident they can restore the backups. I know they can and I don't need reassuring, but they do (well less and less because know they began realizing I really thought this out).
> The chance of someone breaking into your house is sadly much more likely, and them choosing to take any computers they see is almost a certainty at that point.
Got a house break in years ago, they stole no computers.
> What's your next step if you come home tonight and find the house ransacked and the server gone?
Go to the bank, take of one my backup hard drive. Buy a computer, reinstall Proxmox, a VM, Docker CE, redeploy my infra. They still don't have the Yubikeys on my keychain. They still don't have what's on my phone.
Don't think some people here didn't plan for death / theft / etc.
I think a much more likely scenario is an unencrypted drive fails and then what? Do you send it to the landfill unencrypted? Or do you have some process to physically destroy it? Encryption means you can just toss it and feel reasonably confident the data isn't coming back to haunt you.
You should see the hilaribad basis given in affidavits for search warrants that get rubber stamped by judges.
There is no burden of proof and no consequence for perjury. 100% of the search or seizure warrants I have read have had obvious perjury in them.
I encrypt my data at rest not because I fear a burglar breaking in, but because I fear the FBI coming in the front door on some trumped up bullshit. Everyone has a right to privacy, even (and perhaps especially) if they are doing nothing wrong.
I’ve read too many stories of writers and activists getting bogus warrants and charges and arrests thrown at them to inconvenience and harass them to ever have a single unencrypted disk in my house.
I think planning for what happens once you aren't there to manage the setup (whether it be a vacation, hospital stay, or death) is important. It's not something I built specifically to make easy and I should think more on it
The most important thing is to be able to get important data off of it and have access to credentials that facilitate that. You could setup something like Nextcloud to always sync important data onto other people's devices, so make part of that easier
But I think another important aspect is making folks invested in the services. I don't expect my partner to care about or use most of them, but she does know as much as I do about using and automating Home Assistant (the little we've done). Things like that should keep working because of how core they can become to living our lives. It being a separate "appliance" and not a VM will also help manage that
But also that's a lot of hope and guessing. I think sitting down with whoever might be left with it and putting together a detailed plan is critical to any of that being successful
I have given this a lot of thought. I assume the nas and its docker services won't boot starting everything up for someone else. My offsite encrypted backup is probably not recoverable without hiring someone. So:
- I have an ntfs formatted external USB drive to which cron copies over a snapshot of changed daily into a new folder. Stuff like paperless, flat file copy of seafile libraries. The size of that stuff is small <50gb, duplication is cheap. In event of death or dismemberment... that drive needs to be plugged into another machine. There are also seafile whole library copies on our various laptops without the iterative changes. Sync breaks... keep using your laptop.
- I've been meaning to put a small pc/rpi at a friend's place/work with a similar hard drive.
- the email domain is renewed for a decade and is hosted on iCloud for ease of renewal. Although I am not impressed that it bounces emails when storage is full from family member photos which happens regularly so may switch back to migadu.
This applies to so many other things. Who in your house does the taxes? If it's you, would your SO be able to pick up the slack in the event of your death? Can they access all the accounts? Do they even know what all the accounts are? I keep telling myself I need to put together a "what to do if I'm dead" Google doc, but haven't gotten around to it.
I pay $3/mo or whatever for Bitwarden family. It’s wonderful. My wife and I can access all our passwords (and OTP codes!) in one spot. I grouped passwords into folders like “Health” and “Finances”. It has taken us far.
You can put it off indefinitely because nobody anticipates their own death.
You need to add to your threat model having a stroke, where you no longer remember your passwords.
1Password is amazing for this, IMO. My spouse and I have been using 1Password together for more than a decade. One of the first things I set up is a "AAA Read Me First" note with links to a bunch of other notes and documents, including our estate planning stuff.
The biggest thing that makes me stick with 1Password, despite the semi-recent VC shenanigans, is the fact that if for some reason we fall behind on billing (for example, because the credit card got cancelled because I died) the account goes into read only mode forever. As long as 1P is a going concern the data we choose to put there is safe from the biggest risk in our threat model.
Sounds like Stockholm Syndrome to me. If you used KeepassXC, you wouldn't need to worry about this at all since if is entirely free.
I have never used KeepassXC so I looked it up. It seems like it solves a very different use case than what I need:
* shared vault with my spouse's user in our organization account
* multiplatform with sync
* most importantly, available without any of the hardware that I manage being in a usable state
KeepassXC doesn't solve for any of those as far as I can tell.
Just a year before my dad's stroke, my parents documented every account, password, service they had; it was incredibly helpful after he passed with all the stress and chaos
There is a dead man’s switch service [1] which can send an email if you die. In theory if you self host you could trigger something when their email arrives to an inbox you control.
I’ve been thinking of making a version of this that does a webhook but it doesn’t offer a huge amount of value over the email method.
Is the dead man's switch necessary? Unless your homelab contains secrets you don't want revealed until after your death, I'd just put this in a Google doc.
Depends how convoluted your setup is. For some use cases releasing the location of passwords physically written down might help matters or maybe trigger a process to export all data and upload somewhere - somehow.
Seeing some of the discussions around home labs with server racks and k8s doesn’t fill me with confidence that for a majority of use cases a family member would be able to get the data if needed.
> what will happen when I die
Once a year I write couple of DVD with photos. That kind of archaic but easy to understand and reason about media.
Once in a year or two I print some photos in print shop.
I'm glad to see this comment here. People build these projects for family and friends - which is great - and encourage their use, without considering what happens if the only sysadmin suddenly dies. You wouldn't let one person at work hold all of the keys, so the same should be true for your homelab.
While I haven't given all of my keys to my family, there's a clear route for them to get them, and written instructions how to do so. Along with an overview of the setup and a list of friends and colleagues they can turn to, this is enough for them to get access to everything and then decide if they want to carry on using it, or migrate the data somewhere else.
To be frank, if you die, isn't it much more likely your friends and family will just stop using your homelab setup? They'll switch back from Jellyfin to Netflix, replace the smart light bulbs with regular ones, etc.
Could be important data like family photos and financial information in the system.
Yes, of course. They still need to get the all the photos and documents though.
Curious about your setup. Is it extremely unmanageable or have you gone out of your way to make it so?
Unifi network; small proxmox vms for core services; big truenas box for movies, storage, "apps ecosystem" stuff like minecraft servers; baremetal 12 node k8s cluster on opi5s for "research" (coz I do lots of k8s at work).
Each "stage" above is like incremental failure domains, unifi only keeps internet working, core vms add functionality (like unifi mgmt, rancher, etc), truenas is for "fun extras" etc. k8s lab has nothing I need to keep on it because distributed storage operators are still kind of explodey.
Like each part makes sense individually but when I look at the whole thing I start to question my mental health.
Let's explore the implied argument a lil:
Imagine simplest possible deployment you've cooked up.
Now imagine explaining your mother how to maintain it after you're dead and she needs to access the files on the service you setup.
usually, selfhosting is not particularly hard. It's just conceptually way beyond what the average joe is able to do. (Not because they're not smart enough, but simply because they never learned to and will not learn now because they don't want to form that skill set. And I'm not hating on boomers, you can make the same argument with your hypothetical kids or spouse. The parents are just an easy placeholder because you're biologically required to have them, which isn't the case for any other familial relationship)
why does it have to be a non-technical next of kin ? Write down the details for a technically inclined person to follow, maybe a specific friend. Print at the top of the page “show this to X”. In the document explain how to recover the necessary data and replace the setup with a standard one.
I assume most people know at least one person who would do this for them , in the event of their death?
Your assumptions is wrong. Don't assume, verify. Assumptions are the source of many evils.
How do you know? Did you verify or did you just assume?
A good excercise when I had my will written was to create a document describing exactly the case of what happens when I die.
It's probably worth it for most people to go through the excercise
I think the pocket railways are a lot more comprehensible than my local network setup.
> I think about it in terms of what will happen when I die and what I will be inflicting on my family as the ridiculous setups stop working.
I run proxmox too and I've now got a nice little infra at home.
For my family it's simple: I explained them that the infra doesn't matter. The only thing that matter is data. And that there are many, many, many redundant backups of the data and that the backups are functional. That the data are correct (not a single bit missing) and pristine (deduplicated etc.).
Basically 20 years of family pictures and family movies, notarized documents, many proofs of big money transfer (as we now live in an hellhole of KYC/AML where I constantly need to prove money transfer, even when they're so old my banks don't allow me to get that info anymore), all the invoices related to real estate, medical stuff, cars, etc.
My backup procedure uses an intermediary steps that restore the data from the backup and verifies that the data is correct: once that step passes, the backup gets the greenlight. 3-2-1. Even more than 3-2-1.
Cryptographic hashes everywhere, including in filenames: I've got scripts that do verify x% of the files, random sampling style. I'm 100% guaranteed that at least 99.999% of the files are correct. And there are so many backups (online, offline, onsite, offsite, ...), all checksumed. My family won't lose our data. They're literally in various safe and in several countries.
Once I won't be there, they'll have the data up to that point. FWIW both my wife, daughter and brother --although they're not techies-- all happen to be familiar with computers and all took Python lessons. They know what a CLI is.
So even should they have a problem hooking up a hard disk, there's not a world in which they cannot do that:
"LLM> Dear AI, I've got disks with backups of family pictures and notarized documents, how can I access them?"
The world where you couldn't ask that question doesn't exist anymore.
Data of new memories shall be theirs to do deal with though.
P.S: besides that I think homelab'ing is also for convenience and understanding how things (like the network and servers) do work. It's to me more about thinkering and learning than "controlling".
I have been getting into this too. I caution anybody with self-hosting/tinkering tendencies against starting a tech company because it just makes it so much easier to justify this stuff...
Eventually serving a regular old container doesn't cut it anymore and you find yourself needing to pay these weird newspapers nobody reads to publish your business' alias because it's a requirement for a legal DBA which ASIN needs to let you get your own IPV6 block, which you need to truly own you and your customers' IPs and it's not worth becoming an AS without it, but then you can actually move towards physically become your own ISP and then...
The ingress problem people solve with tailscale is one of the hardest. I'm curious to see if it's possible to implement STUN/TURN [0-1] with a generally good mechanism for exposing the server to the Internet by caching all static files and blocking dynamic access to the backend with a loginwall, which authenticates allowed users with email "magic links" -> nonrenewable access tokens. In theory it should not be excessively difficult, expensive, or risky to do this.
It's just relevant enough to what we're doing with remote development environments for me to justify another rabbit hole
[0] https://en.wikipedia.org/wiki/Traversal_Using_Relays_around_...
[1] https://en.wikipedia.org/wiki/STUN
I have ingress set up with Fly.io.
Simple caching nginx config on the remote end with a Fly Wireguard peer set up as an extra container in the appropriate ingress pod.
It's not free but it's the least expensive way I can find to get anycast ingress and not expose any ports to the internet from my homelab.
Mine is much more barebone:
- one single machine - nginx proxy - many services on the same machine; some are internal, some are supposed to be public, are all accessible via the web! - internal ones have a humongous large password for HTTP basic auth that I store in an external password manager (firefox built in one) - public ones are either public or have google oauth
I coded all of them from scratch as that's the point of what I'm doing with homelabbing. You want images? browsers can read them. Videos? Browsers can play them.
The hard part is the backend for me. The frontend is very much "90s html".
HTTP sends password in cleartext. Better to use a self signed certificate at least.
Nice! I have a friend who is starting to program his infrastructure/services from scratch. It's a neat way to learn and make things fit well for your own needs
Impressive setup, but I would assume it to be very operations-intensive because of the high number of deployed components and their complex configuration. Plus, if you are serious about self-hosting, you would need the facilities and infrastructure to deploy it: server rack, redundant power supply, smoke detectors, fire extinguisher... I would never let my PC-grade hardware run unsupervised in my home. And if I understood correctly, you would still have to have some server on the Internet for running your Headscale VPN, so you need your own dedicated Internet connection - ADSL, dial-up, cable modem would not be enough.
Great write-up. I have been tinkering with Immich over the last few months and go back and forth whether I want to just limit it to Tailscale for use away from home or if I want to go through the trouble of setting up a reverse proxy on a VPS. I think my biggest concern is finding a relatively user-friendly monitoring/security solution to alert me if anybody is trying some sort of attack against the VPS
that's a great concern as I'm in the same boat. What have you find with regards to the monitoring/security solution?
I wish I had the time to do any of this. I could probably do it on a weekend but maintaining it, upgrading it to keep up with new releases would be something I wouldn't have time for.
I end up just paying a cloud provider and forget about it.
Anyone else on the same boat? What has been your approach?
Honestly, I self-host about a dozen services and upgrades take me less than a minute per month usually.
I simply have one folder per service, each folder contains a docker-compose stack and a storage directory. Updating is simply a matter of running `docker compose pull` and `docker compose up -d`. Nothing more.
Breaking updates requiring to tweak the config are very uncommon, and even when they happen it's only a few minutes of checking the updated config and applying it.
IMO this is the simplest way to self-host. No VM, no complex software install, nothing more than a simple Docker Compose setup that's fully automated.
That sounds similar to my setup, but each folder is a btrfs subvolume and my update script takes a snapshot before updating. I keep the Docker compose file together with the volumes in that subvolume.
If something breaks I can decide to figure out why, or revert.
With previous setups, I was certainly guilt of not upgrading and doing the maintenance needed. That's one reason why I like using NixOS and ZFS. Both provide really easy rollback options. So all I need to do is run an update and rebuild. If things work, no more for me to do. If things don't, I can try debugging or just revert back to the previous release till I have time to
But also I think using a cloud provider is fine if you're happy with the experience. It is a time sink to get things setup and it's not zero maintenance time. It's reasonable to weight those costs
It's usually not a single weekend. If you're like me, it starts out with thinking it'd be nice to install Plex on an old gaming PC. A year later, it has organically grown into a rube goldberg machine of proxmox and home automation. Which I guess just reinforces your point.
Joking aside, a minimal setup just using docker compose is pretty manageable. Self hosting many projects is as easy as 'docker compose up -d', and upgrades are straightforward as others have pointed out.
I like it. Why Flame, though? It’s built using node, react, redux… meaning you are bringing dozens (if not hundreds) of third party dependencies to your secure kingdom. I don’t think it’s worth it for the start page (could easily be a single html page with hardcoded links)
It's entirely because I've used it before. I just wanted something familiar to solve a problem quickly. I also think it looks nice. I'm not too worried about the security implications, since it is behind Tailscale and Authelia. I'm not committed to it, and do want to explore other options in the future
Why bother with SSO? If your family and closest friends use something like a wireguard client (iOS for example has a very good one that takes only a minute to configure permanently), the users simply switch a toggle and they are now on your private network and don’t need to SSO to anything (provided you have left everything open).
For a small home network the pros of that approach vastly exceed the cons.
The services we use, like Nextcloud or Mealie, are designed for folks to have their own user accounts. SSO means they can use the same login across all of them without me having to manage that for them (and also avoids me having to know their passwords). It does complicated the setup, but not the operation, and that makes it more likely folks will use the services
I self host about 20 separate apps. I’m in the middle of an SSO implementation project because I do not want to continue managing credentials for 20 separate apps.
I’ve considered opening some of these apps to family members, and having one place to deal with any auth issues is a high priority for me.
I can’t agree with your conclusion.
That’s ok. But step back further. Do you need to fine grain permission people on the majority of those apps? If you don’t, then SSO is more of a pain than it’s worth. Simply control the network access and leave the apps alone.
How are you defining fine grained permissions?
All I care about is having separate accounts for each person who will log in, even if I’m the only person.
I again can’t agree with your conclusion that this is more pain than it’s worth. But it’s possible we just have different priorities.
Nice writeup, thank you. I already thought about having NixOS on my server, but currently I prefer proxmox. There are projects with NixOS + Proxmox, but I did not test it yet.
> My main storage setup is pretty simple. It a ZFS pool with four 10TB hard drives in a RAIDZ2 data vdev with an additional 256GB SDD as a cache vdev. That means two hard drives can die without me loosing that data. That gives me ~19TB of usable storage, which I’m currently using less than 10% of. Leaving plenty of room to grow.
I would question this when buying a new system and not having a bunch of disks laying around... having a RAID-Z2 with four 10GB disks offers the same space as a RAID1 with two 20GB disks. Since you don't need the space NOW, you could even go RAID1 with two 10TB disks and grow it by replacing it with two 20TB as soon as you need more. This in my opinion would be more cost effective, since you only need to replace 2 disks instead of 4 to grow. This would take less time and since prices per TB are probably getting lower over time, it could also save you a ton of money. I would also say that the ability of losing 2 disks won't save you from having a backup somewhere...
Oh yeah, I don't think the way I went about it was necessarily the most cost effective. I bought half of them on sale one year, didn't get around to setting things up, then bought the other two a year later on another sale once I finally got my server put together. I got them before I had my current plan in place. At one point I thought about having more services in a Kubernets cluster or something, but dropped that idea
Also agree, RAID isn't a replacement for a backup. I have all my important data on my desktop and laptop with plans for a dedicated backup server in the future. RAID does give you more breathing room if things go wrong, and I decided that was worth it
I went through the same situation and noticed that modern hard drives are big enough for fitting RAID1 into nearly every homelab use case except high-res Video footage (including a bluray movie collection)
Two drives are easy to replace, easy to spare, consume less power and are quieter than 4+.
The only advantage i See in raid5/6 is on 25Tb of storage requirement within 3 years.
As OP says, I think this is the kind of thing that needs to be considered whenever the decision is made.
As another data point, my NAS runs 4x4TB drives. When I bought them new some 2-3 years ago, all at the same time, they were cheaper than buying the equivalent 2x8TB.
My situation was somewhat different, though, since I'm running raidz1. But I did consider running a mirror, specifically in order to ease upgrading the capacity. However, I didn't expect to fill them /that/ quickly and I was right: yesterday it was still less than 70% full.
You are right, but to be max cost effective you could have gone 2 * 4tb or 2 * 6tb for 18 months and then sell the drives still in warranty to upgrade to more storage...
Estimating storage growth is hard but when you monitor it regularly, its saving you much money
Maybe... When I was younger, I used to buy and sell computer stuff. Didn't have much money, so it kinda made sense. But it required me to keep up to date with prices, specs, figure what's the best value, follow markets and jump on an occasion, etc. It got old after a while. There's also value in getting something that just works for you, if it's not absurdly expensive, and forget about it. Do something else with my time.
I still love to tinker and set up a homelab and whatnot, but I don't care that much about hardware anymore. For my needs, if it's at least a 6th gen Intel and I can't hear it in my living room, it's good enough. The NAS lives in my parents' basement, so it can be somewhat louder (with 4 drives instead of two).
For this particular setup, my initial usage was above 4 TB, so I should have went with 2x6, which was /maybe/ cheaper (don't remember), but then it would have required me to deal with selling used gear and go through the motions of upgrading again. Doing this every 4-5 years? Sure. Every year? Hell no.
I've been tempted to use NixOS for my self hosted setup but I just can't bring myself to do it.
My setup is quite simple, it's just a few VMs with one docker compose file for each. I have an ansible playbook that copies the docker compose files across and that's it. There's really nothing more to it then that, and maintenance is just upgrading the OS (Fedora Server) once the version reaches EOL. I tend to stay 1 version behind the release cycle so upgrade whenever that gets bumped.
I do use nix-darwin on my macs so I do _see_ the value of using a nix configuration, but I find it difficult to see if the effort in porting my setup to Nix is worth it in the long run, configuration files don't get written in a short time. Maybe LLMs could speed this up, but I just don't have it in me right now to make that leap
> I've been tempted to use NixOS for my self hosted setup but I just can't bring myself to do it.
I recently tried nixos and after spending a week trying it out, I switched my home network and 2 production servers to nixos. It has been running as expected for 3,4 months now and I LOVE it. Migrating the servers was way easier than the workstations. My homeserver was setup in a few hours.
I also recently bought a jetson orin nano to play and learn on and I set up nixos with jetpack-nixos there too. I know with gentoo this would have been a (much) more painful process.
I have used gentoo for over 20 years and have always felt very much at home. What annoyed me was that the compile times on older computers were simply unbearable. Compiling GHC on my 2019 dell xps just takes 6 hours or something like that.
The big difference for me was NixOS provides really simple rollbacks if something goes wrong, where with Ansible and compose files, that's possible, but you have to do it yourself
But also if you're setup is working for you, I think that's great! It sounds like you have a good system in place
Explored it a bit, but found the incremental gain to be not massive if you're already using IaC of some sort
LDAP author here. I'm happy that LLDAP is mentioned and yet that it is not highlighted. The goal of the project was to have a simple LDAP server that is easy to install/manage for self-hosters without knowledge of LDAP required. Cheers and congrats on your setup!
Thank you for the work and the kind words! I've had a great experience with LLDAP. Really appreciate it
> Authelia provides authentication and authorization for services in a couple of ways. For services that support OpenID Connect it provides a very simple single sign on experience. If not, Authelia can integrate with my reverse proxy (nginx) and require the user login before the reverse proxy allows access to a service.
Recently I found out Gitea or Forgejo can act as an Oauth provider. And since these support ldap you can for example deploy a Samba AD and set it up as an authentication source for Gitea/Forgejo. If you enable the OAuth feature you can connect stuff like grafana and log in with your Samba AD credentials.
To me this is more convenient than running a dedicated auth service considering Forgejo can also provide git, wiki, docker registry (also authenticated) and other function. It's such an underrated piece of software and uses so few resources.
Great blog post, but unfortunately from my experience with my kinda tech-friendly family, i can tell you that not exposing service publicly is an absolute UX killer.
Nobody uses the local nextcloud because they just don't think they can rely on it, it doesn't always work from their perspective, and is too finicky to use, because it needs an external app (Tailscale).
This can be only fixed when the app itself can trigger a vpn connection, and I don't think this is going to happen any time soon.
An easy solve for this is to buy public domains for the sites you want to use, run a static website on them that says “turn on tailscale to access this site, set that up here (link to download a client preconfigured for my tailnet, invite only of course)”, then use tailscale DNS overrides to set up CNAMEs for that public DNS’s (sub)domains which point to the tailnet internal service domains.
This would work if they didn't use the apps, which in the case of nextcloud they do.
I do have to sit down and walk folks through setting up Tailscale, Nextcloud, etc on their devices. So far though, I haven't had any complaints once that is done. Nextcloud just syncs in the background and they can navigate to sites like normal. But my family is probably more tech literate than most, so that helps
Yeah but that means they have to be aware of the need to activate tailscale on their phones manually everytime they want to use your apps.
On PC I agree, you can just leave it running, on mobile though it chews through the battery like it's nothing.
This is kinda similar to something I'm trying to setup. I have most of my self-hosted infrastructure running in docker containers, but I want to put some stuff on a nixOS ec2 instance. Mostly services I want to never go down or be affected by my local network (uptime kuma) and chat stuff (irc bouncer, conduit, soju, etc etc).
I use nixOS on my laptop but don't make many nix projects, and TBH I have no idea how to test this setup locally before deploying it. I have some nix stuff setup that spins up a VM and exposes the ports on localhost, but it's brittle and rapidly spaghettifying. Do you have any tips for testing this stuff as part of a local project?
I've done two kinds of testing
On my NixOS laptop I you can setup services I'm interested in trying, but just run them locally. So I don't setup things like SSL (you can, it sometimes just makes getting a new SSL cert for that same domain take some time). I just update my /etc/hosts to the local IP and can give that a go
For trying out the more complicated setup parts, like SSL, Tailscale, etc, I created a NixOS VM that I setup the same way I wanted for my "production" use case. Once I have the config file the way I wanted, it's as simple as moving it to my non test VM (baring previous mentioned SSL issues). And I only tested one part at a time, adding them together as I went
But also, one of the great things about NixOS is it's really easy to incrementally try things and rollback. Once I got the skeleton of the setup working, I've mostly done my testing on my "production" server without issue
I am curious what are some good enough cheapskate self-hosting setups?
I want to self-host one of those floss Pocket replacements but I don't want to pay more than what these projects charge for hosting the software themselves (~$5). I am also considering self-hosting n8n. I don't have any sophisticated requirements. If it was possible I would host it from my phone with a backup to Google Drive.
Any of the 1L PCs from Dell, HP, or Lenovo. They sip power (5~10 watts), and take up minimal space. I've got a 6 or 7 VMs running on one, and it barely breaks 5% CPU usage.
See https://www.servethehome.com/introducing-project-tinyminimic... for a good list of reviews.
This, when I was a student and had to live frugal (2001-2008 or so), I got a second-hand Dell, put it on top of a high cupboard in my dorm room, and installed a bunch of services (e.g. Trac was very popular in the day for hosting projects).
It won't give you 99.999% uptime, but for that stage in my life it was just stellar. I even had an open source project (Slackware fork) where I collaborated with someone else through that little machine.
Second-hand hardware is also a great way to get high-quality enterprise hardware. E.g. during the same time period I had a Dell workstation with two Xeon CPUs (not multi-core, my first SMP machine) and Rambus DRAM (very expensive, but the seller maxed it out).
Seconded. A dell optiplex micro or hp pro desk with 7th Gen or 8th Gen i5 is approx $40-55 on eBay if you look. Works flawlessly.
Agree. If low cost and maximum value is you're goal, grab a used one of these or similar speed laptop (and you sort of get battery back up in that case)
Really, any machine from the last decade will be enough, so if you or someone you know have something lying around, go use that
The two main points to keep in mind are power draw (older things are usually going to be worse here) and storage expandability options (you may not need much storage for your use case though). Worse case you can plug in a USB external drive, but bare in mind that USB connection might be a little flaky
I would look up intel N100 mini PCs. Extremely low power and fast enough (it’s even got hardware decoding).
I built a NAS with N100 and am pretty happy with it. Price to performance is really good. It runs several services, no issues yet
Which board did you get?
ASUS PRIME N100I-D D4; put 16 GB of RAM in there, but that was probably an overkill
I'm personally happy with my mini-pc+tailscale which is quite cheap, although if it's just for one service and $5/month I don't think it's worthwhile
As a former firefox pocket user, what are the replacements?
I've looked into Wallabag but perhaps there are more I don't know?
Used NUCs, Raspberry Pi / pi zero.
Any old PC with low idle power draw.
I don't have a very complex setup but I've been super happy with gokrazy for my rpis:
https://gokrazy.org/
OS upgrades are easy now and it's declarative but I don't have to learn Nix
Me too. It works really well now that I have a silent version with SSD and passive heat sink. This avoids my son turning it off because the fan noise annoys him. I am thinking about adding kubernetes for failure resilience but that is a work in progress.
I am happy to start digging into Authelia.
Are you using the gokrazy router as well?
I'm not using the router, but I do like reading about Stapelberg's quest for ridiculous home Internet speeds! I make do with standard gigabit fiber haha
Curious what the driver for nixos and packages over docker was. Docker was the huge step up for me in making the homelab easy to run, update, and recover when things failed. It also made managing services endpoints and ports remote easier than when they all lived on the operating system. Wish this was delved into a little more in the post.
I can touch on it more. Docker and compose files are great for getting things going, contained, and keeping everything declarative
But I found the more services I used with Docker, the more time it took to update. I didn't want to just update to latest, I wanted to update to specific version, for better rollback. That meant manually checking and updating every single service, bringing each file down, and then back up. It's not entirely unmanageable, but it became enough friction I wasn't updating things consistently. And yes, I could have automated some of that, but never got around to it
NixOS, in addition to the things I mention in the post, is just a two step process to update everything (`nix flake update` and `nixos-rebuild`). That makes updating my OS and every package/service super easy. And provides built in rollback if it fails. Plus I can configure things like my firewall and other security things in NixOS with the same config I do everything else
Also, Nix packages/services provides a lot of the "containerization" benefits. It's reproducible. It doesn't have dependency problems (see this for morehttps://nixos.org/guides/how-nix-works/). And most services use separate users with distinct permissions, giving pretty good security.
It's not that Docker can't do those things. It's that Nix does those things in a way that work really well with how I think
A pain point you mention is that everyone must run the tailscale client. Have you considered exposing everything on the public internet using something like Cloudflare Tunnels? You can have cloudflare handle auth on their edge network, which mitigates the worry about having to deal with 0-days on a self hosted auth solution. You have a pretty sophisticated directory setup tho so I'm not sure how well this would fit in with the existing infra.
It is something I considered. Ultimately I didn't want to depend on Clouflare (or any other provider) for something as core to my setup as my remote access
But it's a totally valid option, just not one that fit with my preferences
Is there home lab for isolated LAN and "self-sufficient" devices?
I want to have a block of gunk on the LAN, and to connect devices to the LAN and be able to seamlessly copy that block to them.
Bonus: any gunk I bring home gets added to the block.
First part works with navidrome: I just connect through the LAN to my phone with amperfy and check the box to cache the songs. Now my song gunk is sync'd to the phone before I leave home.
This obviously would fit a different mindset. Author has a setup optimized for maximum conceivable gunk, whereas mine would need to be limited to the maximum gunk you'd want to have on the smallest device. (But I do like that constraint.)
I woke up today with a plan of making my DNS at a separate site work properly with ipv6, over my wireguard. I use ULAs for the point to point wireguard link, and GUAs don't like routing to ULAs. I figured the choice was between routing my two sites GUAs over the wireguard when talking to each other, or deploy ULAs in my networks. 4hrs later I had everything set up with ULAs. Had lunch. Decided that was awful. 3hrs after that I've got my GUAs going over the wireguard.
Homelabbing is fun :')
Yes it is, rock on!
> Here is a diagram of where I’ve ended up:
In case the author is around: On mobile (Chrome on Android) the screenshot is not readable at all and there is also no way to open an enlarged version, let alone zoom into the page.
Same on Firefox. Here's the diagram as a zoomable image https://codecaptured.com/blog/images/ultimate-self-hosting/d...
Oh thanks for pointing it out! I've updated it so clicking on the diagram opens it up directly
I too use LLDAP and Authelia. I use Caddy (no Traefik) as a reverse proxy to protect my services using 2FA SSO. It's very easy to use and I can access all my services anywhere in the world without bothering with a VPN.
It's not open-source or self-hosted but putting it out there: CloudFlare zero-trust is amazing and free. In my setup, I have a cloudflared tunnel configured in my homelab machine and I expose individual services without a VPN or opening up my firewall. You can also set up authentication with SSO, and it happens before reaching the backend application which makes it more secure. This is easy for family and friends to use, because they don't need to setup anything from their side, just go to the URL and login. https://developers.cloudflare.com/cloudflare-one/connections...
I seriously don’t understand why would people choose this over not exposing anything at all, except for Wireguard port. I have my client to automatically connect my home LAN when I’m not on WiFi and get access to all my self-hosted services without risking anything. You rely on third party solution which may or may not be made to government agencies. You also need to trust they Cloudflare doesn’t make mistakes, either.
Also, how do you configure Cloudflare for a road warrior setup? How do you track ever changing dynamic IPs? As mentioned, all I need is a Wireguard client and I’m golden.
> You rely on third party solution which may or may not be made to government agencies.
That's a fair point, but for my use case, I feel comfortable enough with CloudFlare given the trade-offs.
> You also need to trust they Cloudflare doesn’t make mistakes, either.
I think the chances of CloudFlare making a mistake are much lower than me or any other individual Developer.
> Cloudflare for a road warrior setup? How do you track ever changing dynamic IPs?
I think you need to read the docs. All of that works without any extra config when using tunnels.
CloudFlare zero-trust is very good, but i thought you need to have Cloudflare as man-in-the-middle on your domain to have this authentication flow work? ie. the TLS certs needs to live with Cloudflare.
Yeah, that is how I use it. You can technically host any TCP including end to end encrypted data through CloudFlare tunnels but you need the cloudflared app installed on the client side to access it (SSO still works even for this case). I find having to manage certificates and installing cloudflared everywhere is too much of a hassle. I understand that proxing through CloudFlare gives them a lot of visibility and control, but I find that risk acceptable given my application.
I’ve got to appreciate putting the matrix server on Coruscant if nothing else :)
Thank you! The naming add a little bit of extra fun to it
Setup your own WireGuard rather than Tailscale.. this is too much like Authy delegating AAA to a third-party.
- Store your SSH public keys and host keys in LDAP.
- Use real Solaris ZFS that works well or stick with mdraid10+XFS, and/or use Ceph. ZoL bit me by creating unmountable volumes and offering zero support when their stuff borked.
- Application-notified, quiesced backups to some other nearline box.
- Do not give all things internet access.
- Have a pair (or a few) bastion jumpboxes, preferably one of the BSDs like OpenBSD. WG and SSH+Yubikey as the only ways inside, both protected by SPA port knocking.
- Divy up hardware with a type 1 hypervisor and run kubernetes inside guests in those.
- Standardize as much as possible.
- Use configuration and infrastructure management tools checked into git. If it ain't automated, it's just a big ball of mud no one know how to recreate.
- Have extra infrastructure capacity for testing and failure hot replacements.
Annoying thing about WireGuard is their outdated and buggy iOS client. When you set up a dns with A and AAAA it'll prefer the A address, even when you're on a 646xlat network, so now that connection is proxied and will time out after a while.
Yep. Other reasons I had to go for IPv4 only a while despite everything else being dual stack. "Argh!" at that one vendor who can't get their act together.
How can one run vanilla wireguard and leverage features offered by headscale? At minimum, a bunch of bash scripts would do the exact same thing, if not worse
Don't do it with bash. You can at least use Ruby, Python to make an API for it, or use configuration management. They really didn't think about being (local) runtime configurable for the dev/ops UX being too ultra *NIX purist with single file plain text configuration. At least it could have a plain text watch directory like daemontools for dynamic reconfiguration.
Headscale already has a clean API in go, why recreate the wheel? For fun, sure but production use, I am gonna stick with it
im using proxmox but struggling to setup subnets and vms
should I be using terraform and ansible?
im using cursor to ssh and it constantly needs to run commands to get "state" of the setup.
basically im trying to do what I used to do on AWS: setup VMs on private network talking to each other with one gateway dedicated to internet connection but this is proving to be extremely difficult with the bash scripts generated by cursor
if anyone can help me continue my journey with self hosting instead of relying on AWS that would be great
> im using proxmox but struggling to setup subnets and vms
That is a pretty broad target. I would say start by setting up an opnsense vm, from there you can do very little to start, just lock down your network so you can work in peace. But it can control your subnet traffic, host your tailscale, dchp server, and adguard home, etc.
As somebody who was quite used to hosting my own servers, before I first set up my homelab I thought proxmox would be the heart of it. Actually opnsense is the heart of the network, proxmox is much more in the background.
I think proxmox + opnsense is great tech and you should not be adding in terraform and ansible, but I am not sure that using cursor is helping you. You need a really good grasp of what is going on if your entire digital life is going to be controlled centrally. I would lean heavily on the proxmox tutorials and forums, and even more on the opnsense tutorials and forums. Using cursor for less important things afterwards, or to clarify a fine point every once in a while would make more sense.
I've found a lot of docs (Proxmox and TrueNAS are both guilty of this) assume you have existing domain or tool knowledge. I'd recommend checking out some videos from selfhosting YouTubers. They often explain more about what's actually happening than just what buttons to select
Also, I found TrueNAS's interface a little more understandable. If Proxmox isn't jiving with you, you could give that a try
I agree Proxmox default networking is lacking/insufficient at best. If you have VLANs or want to do LACP, anything more advanced than a simple interface you'll run into the limitations of the Proxmox implementation quite quickly.
I think the networking experience for hosts is one of the worst things about Proxmox.
Handle subnets on your router, then in ProxMox make the primary network interface you'll be passing to VM's or Containers VLAN aware, with the VLAN tags that it'll support defined and you're good to go.
You don't need any scripts to do that.
Read the docs!
https://pve.proxmox.com/wiki/Network_Configuration#_choosing...
Try using Proxmox's web UI to create a Linux Bridge for each subnet, then attach VMs to appropriate bridges and configure a VM with two interfaces as your router between networks.
Do you really need proxmox ? Would some docker not enough ?
How are you securing taris? Where is your local network firewall? Which one are you using?
Why did you go with Nextcloud instead of using something more barebones, for example a restic server?
This article (https://xeiaso.net/blog/paranoid-nixos-2021-07-18/) walks through a lot of the steps I've done on all my NixOS systems
As for Nextcloud vs a restic server, Nextcloud is heavier, but I do benefit from it's extra features (like Calendar and Contact management) as well as use a couple of apps (Memories for photos is quite nice). Plus it's much more family friendly, which was a core requirement for my setup
As you said, everybody will have their own approach. For me, setting up a dedicated firewall to protect the soft underbelly of my less tech savvy family computers was the primary motivation for starting a home lab. I could not imagine a network architecture without it.
> general approach, lists nixos first
*Leaves page* can't do it...
Very interesting write-up!
At this rate if I keep seeing good article about NixOS I might actually switch for real haha!
Excellent write up. Can I ask why did you choose headscale instead of WireGuard?
I found the Tailscale client experience is quite nice and headscale had built in OIDC support (so easy auth for my users)
If I started this setup later I might have also used pangolin, which also provides a nice management interface on top of WireGuard https://github.com/fosrl/pangolin
Great read. Thanks for sharing.
It’s a shame he doesn’t self host an internal mail server at least with restricted outbound/smtp.
Something like this is very easy to setup with projects such as stalwart which also offers CardDAV and CardDAV (think easy synchronization of calendar and contacts without relying on "cloud").
He already has tailscale + headscale, adding in an internal only mail/collaboration server would be a win.
Hey, I ruled out a mail server for external, since I've heard many people have issues with other providers (Gmail, Outlook, etc) randomly blocking email. Didn't feel I could rely on it
Having an internal only mail server for notifications is an interesting idea. I've been using ntfy and Matrix to achieve something like that, but not all services support those notification methods. I'll keep that in mind!
What’s the power consumption?
That is a great question I don't actually know the answer to. I need to grab something to track it
It's nice to see a home lab on HN. Hardware has become a lost art for many.
If you dont have a home lab, start one. Grab a 1l pc off of ebay. Think center m720q or m920q with an i5 is a great place to start. It will cost you less than 200 bucks and if you want to turn it into a NAS or an Opnsense box later you can.
When it arrives toss Proxmox on it and get your toys from the community scripts section... it will let you get set up on 'easy mode'. Fair warning, having a home lab is an addiction, and will change how you look at development if you get into it deeply.
I credit homelabbing through my twenties with just about everything good that's happened to me in my career. I certainly didn't end up being moderately employable because I'm smart, charismatic, incisive, creative, lucky, educated, diligent, connected, handsome, sanitary, interesting, or thoughtful; no, it's because I have a tendency toward obsession, delusions of grandeur, and absolutely terrible impulse control.
So I started buying junk on eBay and trying to connect it together and make it do things, and the more frustrated I got, the less able I was to think about literally anything else, and I'd spend all night poking around on Sourceforge or random phpBBs trying to get the damn things to compile or communicate or tftp boot or whatever I wanted them to do.
The only problem was eventually I got good enough that I actually _could_ keep the thing running and my wife and kid and I started putting good stuff on my computers, like movies and TV shows and music and pictures and it started to actually be a big deal when I blew something up. Like, it wasn't just that I felt like a failure, but that I felt like a failure AND my kid couldn't watch Avatar and that's literally all he wanted to watch.
So now I have two homelabs, one that keeps my family happy and one that's basically the Cato to my Clouseau, a sort of infrastructural nemesis that will just actually try to kill me. Y'know, for fulfillment.
Not sure if it happens to most, but I have looped back around to not wanting to play sysadmin at home. Most of the stuff I have running I haven't updated in a awhile, luckily since I own it and it's all internal I don't need to worry about anyone taking away my locally hosted apps. Thank the IT gods for docker compose, and tools like portainer to minimize the amount of fuddling around I have to do.
That, and I've learned the stuff I need to from my homelab. Earlier in my career, setting up a light version of a production network at home was hugely educational and I think it was a large part of why I'm a senior today. But now, I don't need to run all that at home unless I have specific learning objectives. So I keep my home network a lot simpler than it used to be, as a result.
Same, replaced the ISP router with my own and have a single box which has storage and compute for running VMs and NFS and that is it. Last thing I want to be doing on a Friday night is debugging why my home network is broken.
Why people need these overly complicated setups and why do they need to have an access point to reach their "den" from anywhere is beyond me. People and their digital gadget delusion.
Security paranoia, but here are the details of my home lab. WHY? If god forbid someone gets in they could in an instant identify the target...
Read the first paragraph and knew you were gonna talk about Nix.
Lol, yeah. It was a journey to get to it, and a slightly shorter journey to feel comfortable with it, but it has won me over
I love the idea of nix... but i want ubuntu nix or fedora nix :-P
It needs to be stupid easy and reliable.
I think that's a fair point. Kinda like with Arch, you do have to know what you want to setup NixOS right now
I really like what's happening in the ublue space were folks are tweaking and optimizing distros for specific use cases (like bazzite for gaming) and then sharing them
NixOS does support that to an extent, but it certainly doesn't have the same community movement behind it like those
IMO this is too complicated. I think products like the Synology Disk Station strike a better balance between ownership of data and maintenance over time. Tailscale even publishes a client for Synology products.
Everyone will have different goals and preferences. For instance, my dad just wanted a way to backup and remotely access some files, so we got him a Synology NAS. It's great for it's target users and if you're one of them, awesome!
I just don't like the lock-in that you get Synology. Plus I do enjoy tinkering with these things, so I wanted to put together something that balances usability, complexity while minimizing that lock-in
Is this the same Synology that a few years ago numerous people had their boxes ransomwared when they were open to the public internet. Synology continues to be shit value for the tools you get and as much as I want the convenience of ready to go software and hardware they cannot be relied on.
I am thinking about replacing my Synology BUT I have had three and now ECC over the last 20 years and they have done their job faullessly. There are stupid things now like complaining about non Synology ram modules.
The latest news I heard is DSM complaining about non Synology hard drives too.
Outside of the stated requirements because its not fully open source, but https://www.cloudron.io/ made all my self hosting pains go away.