It depends. I’ve ran small websites and other services on a old laptop at home. It can be done. But you need to realize the risks that come with it. If the thing I’m running for fun goes down. someone might be slightly annoyed that the thing isn’t accessible all the time, but it doesn’t harm anyones business. And if someones livelihood is depending on the thing then the stakes are a lot higher and you need to take suitable precautions.

You could of course offload the whole hardware side to amazon/hetzner/microsoft/whoever and run your services on leased hardware which simplifies things a lot, but you still run into a problem where you need to meet more or less arbitary specs for an email server so that Microsoft or Google even accept what you’re sending, you need to have monitoring and staff available to keep things running all the time, plan for backups and other disaster recovery and so on. So it’s “a bit” more than just ‘apt install dovecot postfix apache2’ on a Debian box.

Others have already mentioned about the challenges on the software/management side, but you also need to take into consideration hardware failures, power outages, network outages, acceptable downtime and so on. So, even if you could technically shoehorn all of that into a raspberry pi and run it on a windowsill, and I suppose it would run pretty well, you’ll risk losing all of the data if someone spills some coffee on the thing.

So, if you really insist doing this on your own hardware and maintenance (and want to do it properly), you’d be looking (at least):

• 2 servers for reundancy, preferably 3rd one laying around for a quick swap
• Pretty decent UPS setup, again multiple units for reundancy
• Routers, network hardware, internet uplinks and everything at least duplicated and configured correctly to keep things running
• A separate backup solution, on at least two different physical locations, so a few more servers and their network, power and other stuff taken care of
• Monitoring, alerting system in case of failures, someone being on-call for 24/7

And likely a ton of other stuff I can’t think of right now. So, 10k for hardware, two physical locations and maintenance personnel available all the time. Or you can buy a website hosting (VPS even if you like) for few bucks a month and email service for a 10/month (give or take) and have the services running, backed up and taken care of for far longer than your own hardware lifetime is for a lot cheaper than that hardware alone.

NAS stands for ‘Network Attached Storage’ and there’s dedicated hardware for that task from multiple brands. It’s a somewhat spesific thing and from what I understand you have a multi-purpose server running on your network. For discussion it’s better to use the established terminology to avoid confusion on what’s what. Your generic server can of course act like a NAS, but a 100€ Synlogy NAS can’t (for the most part) act as a generic server.

Similarly there’s a dedicated hardware for routers and they are not the same than generic servers which can run whatever. Dedicated routers do some things way better/faster than generic server, and there’s pretty much always a trade-off between the two. You can of course install hardware to your server to be as good as or even better than any consumer grade router and run a pfsense on virtual machine on top of it, but that’s going to be at least more expensive than dedicated hardware.

So, your server is running pihole in a container on the same network address/hardware than the rest of your server, and I suppose you already gathered from other messages that the firewall component on it treats traffic coming from outside the server itself differently than traffic originating from the server itself. For this spesific case I’d say it’s just simpler to configure the server to use DNS server as localhost:1053 than trying to work out firewall forwarding rules for it, if possible. If not, and you absolutely insist that your pihole runs on a unprivileged port and that your server also has to use pihole as DNS sever, then you need to dig out a firewall config for outgoing traffic which redirects the destination port. Or you could set up a dns proxy on the server which uses pihole as upstream and serves addresses to localhost only or one of the other multiple ways to achieve what you’re after, but each of those have some kind of trade-off and there’s too many to go trough in a single post.

If the firewall was running on a router then you’d need to DNAT back to the same network from which they originated and that is (in general) quite a PITA to get running properly. My understanding is that the firewall doing port forwarding is running on the NAS. And we don’t have much information on what that ‘NAS’ even is, I tend to think devices like qnap or synology when talking on NAS-boxes, but that might as well be a full linux-system just running CIFS/NFS/whatever.

OP could obviously use his router as a DNS server for the network and set upstream DNS server for the router to pihole, but that’s a whole different scenario.

As it’s only single device I’d suggest configuring DNS server for that to <ip-of-nas>:1053. Port forwarding rule on the nas firewall most likely applies only to ‘incoming’ traffic to the nas and as locally generated DNS request isn’t ‘incoming’ (you can think it as ‘incoming’ traffic is everything coming via ethernet cable into the nas) then the port redirection doesn’t trigger as you’re expecting.

Bare metal server sounds like optimal solution for you and set up a hypervisor on top of it, so it’s pretty trivial to migrate VMs to your own hardware when needed. But then for your ‘long term’ environment VPS would most likely be better and migrating a full VM from your hypervisor to VPS is a bit more work, but can be done.

I don’t know about providers in Australia, but Hetzner has both and combined billing and my personal experience with them is pretty good. But I’m in Europe, so bandwidth nor latency is not a problem.

My experiences are few years old, so I don’t remember excact models anymore, but some back-ups models (es series rings a bell, but as I said, it’s been a while) had batteries with soldered connectors and form-factor which (at least at the time) wasn’t available from anyone else than APC.

I have older 1500VA FSP UPS, I don’t think that exact model is available anymore, but it’s been solid for several years. It currently has 3rd or 4th set of batteries and they are standard bulk batteries, so replacements are easy to find from anywhere. Only problem I’ve had with that is that on display it doesn’t give out clear warnings when batteries degrade and it has crashed my system few times in a power outage, but I’ve been lazy and didn’t bother to properly monitor it nor have scheduled battery replacements, so that’s mostly on me.

Eaton seems to be pretty solid too, but I don’t have a ton of experience on any of their models. Local suppliers had dirt cheap PowerWalker UPS’s a few years ago, but one of them didn’t survive when battery died, so maybe I got what I paid for. Those worked fine too, but apparently they cooked the carging circuit when battery degraded.

This is of course just my own experience over a few models, but personally I wouldn’t spend my money on APC. Propietary batteries and multiple failures after battery replacement at work few years back were enough to choose something else.

I recommend Hetzner too. I’ve been a happy customer for a decade. Support, should you need it, works well and services are rock solid.

You can run clonezilla on your shell session, just apt install conezilla (or whatever variant you’re using) and it can do the trick. Dd will almost surely work too, but that leaves a ton of responsibility to you instead of making any sanity checks on the way. That makes dd very powerful tool and it has saved my ass a multiple times, but if you already have a working partitioning schema clonezilla has a ton of options to make your life a lot simpler and a likely a bit faster than dd.

more specific to a subset of people who have time to bother

And that subset of people needs to have at least some kind of mindset to learn the viable minimum skills to even start with and a will to learn more and more and more. I’ve done various kinds of hosting as a career for couple of decades and as things change I’m fighting myself if it’s worth my time and effort to keep my home services running or should I just throw money to google/apple/microsoft/whoever to store my stuff and manage my IOT stuff and throw the hardware into recycling bin.

I have the skill set required for whatever my home network might need up to a point that I could somewhat easily host a small village from my home (money is of course a barrier after a certain point), but I find myself more and more often thinking if it’s worth the effort. My Z-wave setup needs some TLC as something isn’t playing nicely and it causes all kinds of problems with my automations, my wifi network could use a couple of sockets on the walls to work better, I should replace my NVR with something open source to include couple of more cameras around the yard and have better movement recognition and cameras should go to their own VLAN and so on.

Most of that stuff is pretty basic to set up and configure (well, that z-wave network is a bit of it’s own thing to manage) and it would actually be pretty nice to have all the things working as they should and expand on what I have to make my everyday life even more simpler than it already is. But as there’s a ton of things going on in life I just rather spend few hours gaming from my sofa than tinker with something.

That’s of course just me, if you get your reward and enjoyement on your network then good for you. Personally I think I’ll keep various things running around, but right now in this place I’m at, the self hosting, home network and automation and all that is more of a chore than a hobby. And I’m pretty sure I don’t like it.

As far as I know it is the default way of handling multiple DNS servers. I’d guess that at least some of the firmware running around treats them as primary/secondary, but based on my (limited) understanding at least majority of linux/bsd based software uses one or the other more or less randomly without any preference. So, it’s not always like that, but I’d say it’s less comon to treat dns entries with any kind of preference instead of picking one out randomly.

But as there’s a ton of various hardware/firmware around this of course isn’t conclusive, for your spesific case you need to dig out pretty deep to get the actual answer in your situation.

have an additional external DNS server

While I agree with you that additional DNS server is without a question a good thing, on this you need to understand that if you set up two nameservers on your laptop (or whatever) they don’t have any preference. So, if you have a pihole as one nameserver and google on another you will occasionally see ads on things and your pihole gets overrided every now and then.

There’s multiple ways of solving this, but people often seem to have a misinformed idea that the first item on your dns server list would be preferred and that is very much not the case.

Personally I’m running a pihole for my network on a VM and if that’s down for a longer time then I’ll just switch DNS servers from DHCP and reboot my access points (as family hardware is 99% on wifi) and the rest of the family has working internet while I’m working to bring rest of the infrastructure back on line, but that’s just my scenario, yours will most likely be more or less different.

My bank uses 6 digit ‘customer number’ (which is set by the bank) and that’s verified with an app and a personal PIN (app shows ‘login attempt ABCD at mm.dd. hh:mm’ where ABCD is shown on login page too) or via SMS OTP (again with ‘ABCD’ verification). And again with personal pin + app or OTP to confirm transactions. The app itself can be protected with a fingerprint or phone pin and every new installation needs to be registered to the system, so I can’t just use my phone app to access my wifes account (or anyone elses) but I still can map multiple accounts (like corporate ones) to the same installation.

I think that’s pretty reasonable approach.

Back in the day with dial-up internet man pages, readmes and other included documentation was pretty much the only way to learn anything as www was in it’s very early stages. And still ‘man <whatever>’ is way faster than trying to search the same information over the web. Today at the work I needed man page for setfacl (since I still don’t remember every command parameters) and I found out that WSL2 Debian on my office workstation does not have command ‘man’ out of the box and I was more than midly annoyed that I had to search for that.

Of course today it was just a alt+tab to browser, a new tab and a few seconds for results, which most likely consumed enough bandwidth that on dialup it would’ve taken several hours to download, but it was annoying enough that I’ll spend some time at monday to fix this on my laptop.

I mean that the product made in here is not the website and I can well understand that the developer has no interest of spending time for it as it’s not beneficial to the actual project he’s been working with. And I can also understand that he doesn’t want to receive donations from individuals as that would bring in even more work to manage which is time spent off the project. A single sponsor with clearly agreed boundaries is far more simple to manage.

You do realize that man pages don’t live on the internet? The kernel.org one is the offical project website, as far as I know, but the project itself is very much not for the web presense, but for the vastly useful documentation included on your distribution.

https://xkcd.com/2347/

The threat model seems a bit like fearmongering. Sure, if your container gets breached and attacker can (on some occasions) break out of it, it’s a big deal. But how likely that really is? And even if that would happen isn’t the data in the containers far more valuable than the base infrastructure under it on almost all cases?

I’m not arguing against SELinux/AppArmor comparison, SElinux can be more secure, assuming it’s configured properly, but there’s quite a few steps on hardening the system before that. And as others have mentioned, neither of those are really widely adopted and I’d argue that when you design your setup properly from the ground up you really don’t need neither, at least unless the breach happens from some obscure 0-day or other bug.

For the majority of data leaks and other breaches that’s almost never the reason. If your CRM or ecommerce software has a bug (or misconfiguration or a ton of other options) which allows dumping everyones data out of the database, SElinux wouldn’t save you.

Security is hard indeed, but that’s a bit odd corner to look at it from, and it doesn’t have anything to do with Debian or RHEL.

If I had to guess, I’d say that e1000 cards are pretty well supported on every public distribution/kernel they offer without any extra modules, but I don’t have any around to verify it. At least on this ubuntu I don’t find any e1000 related firmware package or anything else, so I’d guess it’s supported out of the box.

For the ifconfig, if you omit ‘-a’ it doesn’t show interfaces that are down, so maybe that’s the obvious you’re missing? It should show up on NetworkManager (or any other graphical tool, as well as nmcli and other cli alternatives), but as you’re going trough the manual route I assume you’re not running any. Mii-tool should pick it up too on command line.

And if it’s not that simple, there seems to be at least something around the internet if you search for ‘NVM cheksum is not valid’ and ‘e1000e’, spesifically related to dell, but I didn’t check that path too deep.