Please wait...
I'd be more interested in running a desktop environment inside an LXC
container.
Can you do that? Yes I know, you're not supposed to do that, but I
wonder if it's possible.
I am sure you can. I am not sure it makes sense.
For what its worth - Reason i had the 'minimal desktop' on the host
was so it was 100% standalone/portable, but with option to use PVE
backup server across VPN. I can use the native browser to get
spice/vnc sessions, remote Xterm, or just ssh, to access desktops in
a vm. and the browser to manage the host. And no, i dont normally do
that. This was an exception case as it needed to be portable and
still have access to my backups, tho i have heard of people doing it
on a regular basis. But i would think virt-manager + kvm would be
better in most cases.
I just have my Proxmox cluster in a dedicated LAN into which I can VPN everytime I need to. The clusters are a bit heavy to carry arround, but I can access my stuff from anywhere so it is nearly as if I was carrying my virtual machines with me.
That said, I have my own DNS and TLS/Certs built for this stuff so I can reach my virtual (and real) machines by name without having to think about it and for having a secure channel even within the VPN (since I don't necessarily trust some of the network gear in te infrastructure).
Normally if i'm not at my house, id use VPN ( or guacamole ) to get to whatever i want.
This as an odd ball edge-use case. Was more relevant about the similar loss of network like docker, than the setup itself :)
I just have my Proxmox cluster in a dedicated LAN into which I can VPN
everytime I need to. The clusters are a bit heavy to carry arround, but
You can carry a cluster around; all you need is a laptop running three virtual machines with nested virt in which you run Proxmox and then build a cluster out of those... hehe
(Well that's how you can test clustering anyway ... if you actually WANT a cluster, then you need three laptops.)
In my case I just built this little server, it ended up on a shelf on the network wall in the garage, and I kind of want to be able to "do computer stuff" in that room. Running a desktop in a partition that I can blow away seems like a better idea than running it in the root.
You can carry a cluster around; all you need is a laptop running three
virtual machines with nested virt in which you run Proxmox and then
build a cluster out of those... hehe
I am sure I could build something with raspberries that would count as a portable cluster.
ewwwww
Thu Mar 14 2024 12:33:39 EDT from darknetuserI am sure I could build something with raspberries
I am sure I could build something with raspberries that would count as
a portable cluster.
Yeah, you could be a lillicluster whore like Jeff Geerling. Or you could buy one computer with enough power to blow away the biggest pi-cluster.
I know, I know, it's all proof of concept stuff. And for learning. And for whoring on YouBoob. In my case I don't care, I have access to a million dollar lab environment at work. My stuff is here to run actual workloads, and I'm going to do that as efficiently as I can.
Myself, if ( or when ) i was playing with cluster stuff to try a few things, id just do nested VMs on a larger host... One cheap xeon would eat a room of PIs and be far easier to setup, break, tear down, rebuild, snapshot, bla bla etc.
Only reason id want to run a bunch of lower power would ARM boards would be for a true fault-tolerant ceph storage cluster. Smallish ARM boards excel for that. ( tho id still not choose pi.. ech ) ok i guess there are 2 reasons.. the other is for the other use of cheap ARM ., embedded IoT.. but im more liable to use ESP32 instead these days.
And, im talking cheap ARM since RPI was brought up. The high end, not throwaway cost stuff, of course i love for 'real' use, as everyone around here knows by now :)
Thu Mar 14 2024 13:21:37 EDT from IGnatius T FoobarI am sure I could build something with raspberries that would count as
a portable cluster.
Yeah, you could be a lillicluster whore like Jeff Geerling. Or you could buy one computer with enough power to blow away the biggest pi-cluster.
I know, I know, it's all proof of concept stuff. And for learning. And for whoring on YouBoob. In my case I don't care, I have access to a million dollar lab environment at work. My stuff is here to run actual workloads, and I'm going to do that as efficiently as I can.
Yeah, you could be a lillicluster whore like Jeff Geerling. Or you
could buy one computer with enough power to blow away the biggest
pi-cluster.
You don't need to be acidic. My point is that a cluster-in-a-laptop is as much cheating as a cluster with cheap SBCs :P
so a bit of frustration with my attempt at wireguard..
use case: now that my off-site backup plan is ruined. i thought about dropping off a tiny ARM box with a couple of TB storage at my brothers house. Have it auto connect to my network, and once a week or so do an RSYNC over ( with z... compress the data )
Trying not to be fancy, 'just enough' to get the job done. Cant ask for ports to be opened nor do i want to effect brothers network at all ( and it might move to another house at any point so has to be self-contained ), so I figured id use either openvpn or wireguard and reach out, not in.
Openvpn - setup is easy, i have a server now for when i travel. But getting things so i can get back into the remote device, not so easy.
Wireguard, got it setup using some 'easy wireguard docker image' ( wg-easy ) that has a gui that creates config files for you, even QR codes. It worked for incoming just fine and was pretty painless. Problem is that its docker, so again, getting back to my device would be a pain. Either screwing with networking in docker, or perhaps a shell in the container, bla bla. But it did at least connect and work.
- Trying it on the bare OS using the below link. I can only get it sort of working.
- Server, starts, runs properly. Client, i can connect manually using wq-quick but not yet as a auto start service ( systemd..bleh, but i can figure that out later once network is right ). But cant seem to get it to be reachable on my network. Only can access the client via the server, using ssh. If what im doing just isn't doable, i guess a double hop via server isn't end of world, just not as convenient ).
- If i follow directions exactly, and put both devices it on an unused subnet, i expect that to happen, but i also lose all connectivity out of the client, other than being able to ping the gateway.. Again, not end of world i as i 'could' use the server and ssh over.. but it should work.. it worked when i had the docker version of server going.
- If i change the IPs to unused addresses on my normal subnet, its pretty much toast.
So i guess the real question is, with the above issues, is he missing some steps? Or am i just a dumbass and missing something obvious?
Link im using -> https://www.stavros.io/posts/how-to-configure-wireguard/
Ok, scratch the above. I tore both machines down, redid them. It works. i donno. Some days i hate computers.
F***ing feh.
The whole family has been getting wifi dropouts ever since I upgraded my Mikrotik h-EX to whatever software was current as of a couple of days ago.
And today I noticed it too. Wifi going in and out all over the place. And my IPv6 is gone. Picks up a prefix and hands addresses out to clients but doesn't forward. And the web UI is also gone.
This Does Not Bode Well. I need to have a ridiculously solid network when I move my hosting stuff back home later this year.
I disabled CapsMan, removed all AP's except the centrally located one, and turned it into a standalone AP. Wifi is better, but not perfect. And yes I checked to make sure it wasn't just that AP crapping out, because I was connected to a different one before.
Wondering how my phone decides the wifi network is out. Maybe it pings the gateway or something. With the wifi down to one standalone AP it's possible to try a different router. I could press the NanoPi into service for that role, since it's built for that sort of thing, but I don't really want to give up my ARM development host. I dunno ... guess I'll watch it for a day or two and figure something out.
i suspect it hits an actual site, as the gateway could be up, but non functional.
Could it be interference and just coincidental timing ?
Thu Apr 04 2024 00:43:51 EDT from IGnatius T Foobar
. Maybe it pings the gateway or something.
I ended up factory-resetting my router which solved most of the problems.
I also kept it with just one access point after realizing that the one in my office can actually cover the whole house without difficulty. Things seem ok now. Wifi is solid and all my local stuff works again. But I still have the problem where my Android loses its IPv6 default gateway after a while.
So that explains the tremors.
Fri Apr 05 2024 16:18:49 EDT from IGnatius T Foobar
I ended up factory-resetting my router which solved most of the problems.
So that explains the tremors.
Ok I just must share this story, since we are talking about networks and earthquakes.
Now you know I normally don't do tech support for anyone who isn't either sleeping with me or signing my paycheck (in other words, my wife and my boss can ask for computer help, everyone else can fuck off). But I did build the network in our church building so I'll answer a call for help there, with the condition that it has to be a network problem -- no computer problems and definitely no printer problems. As we all know, the three words that repulse tech people the most are "I can't print."
But parts of the network were out, and it's my day off, so off I go. It was a weird problem, the core network was good, the wifi was up, but parts of the network were out. After some simple tracing I found the culprit: the earthquake had shaken the power cord loose from one of the switches upstairs.