New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
This is a step ahead of where I am, which is excellent!
Let me think about it. And try it.
The idea of getting something going that would benefit all the users sounds especially good!
Do you want something temporary while I am thinking and trying to get what you want going? Might be a day or so. Or do you prefer to wait?
Massive thank you!
Logged in just fine, will test more ASAP.
I see there are discussions about distributions here; is this node "stuck" with Ubuntu or is it possible to go with Debian?
A Chisinau VPS slot has been assigned to @FrankCastle, so, as of now FOSSVPS still has 2 available dedicated IPv4 slots in in Chisinau ( Thanks Alexhost!
) and 3 available NAT VPS slots in London ( Thanks Onlyservers! 
). Sorry, LA is fully assigned ( Thanks @fmxm! 
).
FOSSVPS has 29 VPSes assigned in LA, London, and Chisinau.
That is very kind of you to offer but I'm in no rush. Take whatever time you need. In the mean time I'm here if you need anything, just let me know.
Hi @zejjnt!
Usually I go for Debian or NetBSD. The Chisinau Node was delivered with Ubuntu. I thought it would be lots of fun to go with Ubuntu since I haven't used it for a while.
You are right that it would be super great for everyone to be able to install whatever open source OS they want. My question is, though, how to arrange universal open source OS install for everyone without using closed source software? Also, there is my own little preference for ssh and command line, much simpler than something like OpenStack.
One idea might be to make netboot.xyz available. I've been talking with @babywhale about adding netboot.xyz to BashVM. Unfortunately, I have almost no experience with netboot.xyz.
Another idea proposed by @FrankCastle is to use limited no-login ssh:
@zejjnt How would you implement open source OS install freedom on our Node? Thanks!
With only the limited ssh access how would you change your OS image?
The boot parameter I provided in the virt-install takes care of that. It provides access to PXE so I can install anything I want using that. No need to attach a cdrom drive or set custom iso images or anything like that. Simple and elegant.
The only real drawback with the virt-install I provided is the "os-variant unknown" section. Going with a specific operating system provides a lot of of optimizations but since I want the flexibility to install whatever that's the best I could come up with. Setting it to a specific linux flavor and then me installing BSD wouldn't make any sense so going with unknown while there can be a little bit of a performance hit it keeps it universally open. I've tried using "os-variant generic" as well but that complains more than unknown. I haven't run any extensive benchmarks between the two. Maybe someone else on here could provide feedback on what they think the best option here would be.
@FrankCastle Excellent! I have to look into this. I don't remember previously seeing
--boot hd,network,menu=on \Thanks for the hint! Much appreciated!
By the way, I'm looking for some small financial contributions to aggregate the approximately $200 required to register FOSSVPS as a non-profit. After registration, I hope to get a 501(c)(3).
For reasons related to the tax exempt status, expenses might best be paid with funds the majority of which derive from contributions by the public. In other words, maybe I shouldn't just pay the $200 myself. I already have $50 contributed, so we are on the way.
If by any chance you know somebody who might want to contribute, please put them in touch with me.
An update from my side: I've been working on a page that would include most of the data sources of interest that the VPS provides. I am still debugging some items, but hope to get it ready soon.
Yesterday and the day before were the first full days with the NTP server in the NTP pool. Traffic continues to be below 1Mbit/s on average, making for about 12GB traffic in+out per day. I expect it to go up a bit further still, but not much. I.e., I don't think it will exceed 1Mbit/s by much, if at all.
There is a slight variation of traffic throughout the day, but like the traffic demand more on the lower end of what I've seen so far.
The above figures are all the traffic of the VPS, but NTP is the dominant one for most of the time. I've throttled the background package downloads so they don't distort the picture too much. The stretches in the figure where there is noticeably more inbound traffic than outbound traffic could still be those downloads. Or it could be some misbehaving clients of well-known types seen elsewhere as well. Those are detected using Miroslav Lichvar's ntpguard tool, and subsequently filtered (they keep hammering a server at such a high rate so as to suggest they don't even use the time they get back with most responses).
The server seems to be doing well in the pool, nothing out of the ordinary in the delay and score graphs.
I've been loosely monitoring, and comparing, the CPU load as reported inside the VPS, and as seen from the host. Even without trafficserver running, the load seen on the host side is still about one CPU core most of the time, while the load seen inside the VPS is mostly less than 30% of a virtual CPU core, with chronyd reaching about 15% for a packet rate of about 1000 packets/second.
I've not yet stopped various services inside the VPS individually (e.g., didn't want to defer the traffic estimate further by even temporarily stopping chronyd) to see whether there is a single one, or a subset, that cause the majority of the load seen on the host side. But I am starting to think that the discrepancy between the load levels inside and outside the VPS might simply be due to various overheads associated with the virtualization layer, with different programs exercising the different types of overheads to differing degrees (based on what they are supposed to do). So I will dig a bit deeper (e.g., stop individual services to see which ones make most of a difference), but I don't think any urgency is needed.
Maybe I should just reinstall your VPS with Debian?
It was part of the blob I provided.
Sure thing, always happy to help. I tried to make it as hands off as possible and provided everything required so all you had to do was cut and paste.
Definitely, will do.
@Starnberg Where do I look for more details of (1) how you are collecting and (2) how you are displaying the data? Or details of a similar setup, not exactly yours, also would be fine. Thanks!
@Not_Oles, the first two graphs are done using vnStati, based on network data collected by vnStat (or rather, the vnstatd daemon responsible for collecting the network data). Both are standard packages under Debian et al. vnStati is for rendering nice graphs, and a sample CGI script included with the package then orchestrates the generation of different types of graphs, and of HTML pages collating those graphs. vnstat is a tool to access the same data from the CLI.
The last two graphs are just included from the NTP Pool side, they are based on data the monitoring infrastructure of the Pool collects (and which is also accessible in raw form).
I am also using darkstat, which unfortunately seems no longer maintained by upstream, but is still a standard package. It provides a slightly different view of network traffic. I like it because one can use a pcap filter expression to select traffic to be counted, e.g., just NTP traffic (vs. vnstatd counting all traffic passing through an interface).
And I've installed a HetrixTools agent, having learned about it in this thread. It is one source for the CPU data from inside the VPS that I mentioned verbally.
@Starnberg Okay thanks for the extra explanation! Very helpful!
@Not_Oles I had been planning on providing a more elaborate description alongside the web page I mentioned that collates output from various tools in one place once I'll share that page. It will also include output from some other tools/data sources.
Thanks to @Not_Oles for working with me to get everything working. Thanks to @alexhost for donating the hardware. Thanks to FOSSVPS for such an awesome project. Thanks to everyone else involved!
@FrankCastle has been teaching me about setting up a VPS configuration where users have access to pxe boot and VNC via an ssh tunnel (without login to the node).
Maybe Frank might post about how he made an OS install following the configuration of his VPS. I haven't yet seen this type of OS install from the user perspective.
I could post about the sever node configuration if anyone is interested.
I just slapped OpenBSD 7.7 on it as a test to make sure everything worked as expected. I can't promise I'll have the time to take screenshots and such when I start messing with Debian 13 but if I do have the spare time and anyone is interested I can definitely try to show the highlights (too lazy to make it super in depth). If anyone is interested or has specific questions about it just let me know.
I just refuse to use templates or any cookie cutter installs because I'm a paranoid tin foil hat wearing conspiracy theory person that has to have full disk encryption and a few other tricks to make it more difficult for anyone to snoop on my little slices. In reality, I doubt anyone actually wants to spy on my activities anyway but better safe than sorry I suppose.
I suppose I oughta just try it from the user side.
Locally, run something like
ssh $user@$IP -L 59XX:localhost:59XXThen in a local VNC client, localhost:59XX. No password needed?
What should I initially expect to see inside the VNC client?
How do I start an install?
Hi @Not_Oles
If i am qualified. I'd like to host nginx, phpmyadmin, mariadb, wordpress etc for my blog
any location and I like to use ubuntu
thanks...
Hi @swarupgolui!
Welcome to LET! Congrats on your new account!
I'd love to have you on one of the FOSSVPS servers!
Maybe you could make it easier for me to say yes by sharing some additional information about yourself and your computer background?
Do you have an existing blog somewhere?
Anything else that you want to share?
Thanks!
Best wishes!
Tom
Hi @Not_Oles
Do you have an existing blog somewhere?
Yes, I had but I miss the renew period because 1 month ago I had a bike accident and got injured badly, rn domain status is pending delete, I already talk with Namecheap but they said not posable to renew rn because domain status is pending delete
they said domain will be available to re-register again on 26/08/2025
Maybe you could make it easier for me to say yes by sharing some additional information about yourself and your computer background?
Nothing special, previously I was using AWS free tier for my domain but rn AWS free tier is no more free..
You should, that's a good idea.
Pretty much, you got it.
You see will most likely see SeaBIOS boot menu or PXE might have stalled out and be sitting at a screen saying it wasn't able to boot. You just reboot to start the loop all over and cath the process this time through to start your install.
Same way as every other provider that doesn't have an install template for the operating system of your choice. Standard PXE install, nothing special at all.
@swarupgolui
Thanks for your message! Sorry to hear about your bike accident! Hope everything is healed soon!
As you can understand, from my perspective of trying to administer the servers, a significant is that you have a brand new LET account. You haven't been part of our community for long enough to use your posts and comments to get a picture of how positive and helpful you are. As you will understand, everything is right about what you are saying, but there is no idea about how anything could be confirmed.
For example, you didn't mention the name of your domain. Maybe your former site still remains available on archive.org?
Until your positive and helpful post and comment history is built, you still can consider providing more information -- if that's a good choice from your perspective.
Thanks again for starting your FOSSVPS application!
Best!
Tom
Thanks to FOSSVPS' wonderful server Node donors: OnlyServers, Alexhost, and @fmxm.
Thanks to @babywhale for BashVM! Thanks to @AuroraZero for hosting the FOSSVPS wesite on YetiNode free!
Hello! Happy Sunday!
The remaining two dedicated IPv4 VPSes in Chisinau have been allocated. Thanks to Alexhost for their kind donation!
More Chisinau dedicated IPv4 VPSes might become available if additional IPv4s are donated, if another server is donated, or if one of the current clients moves elsewhere.
Two of the three remaining London NAT VPSes have requests under consideration, but not yet decided. At least one London NAT VPS seems currently to remain available. Thanks to Onlyservers for their kind donation!
I look forward to further discussions of how things are going on our existing VPS installs, additional server donations, and more FOSSVPS clients!
Best!
Tom
All the Chisinau VPSes have been assigned.
All the London VPSes have been assigned.
All the Los Angeles VPSes have been assigned.
Thanks to FOSSVPS' wonderful server Node donors: OnlyServers, Alexhost, and @fmxm.
Thanks to @babywhale for BashVM! Thanks to @AuroraZero for hosting the FOSSVPS wesite on YetiNode free! 
Thanks to our 34 wonderful users!
More soon!
So anyone with such a user account on the host can get access to the VNC console for any VPS running on that host then?
Hell no, that would be a huge security lapse. Each account is only allowed to their dedicated vnc port. It is a very valid question though but I covered all of these details with Not_Oles in direct messages so not every aspect of the config can be seen in this thread. We aren't trying to give everyone access to every VM.
@cmeerw If I understand right, there seem to be at least three reasons why no user gets access to all the VNC consoles. First, if I understand right, VNC isn't running on the currently deployed VPSes, except for one. Second, each VPS' console would run on a different port, and each user account would get restricted access only to that one port. Third, VNC has password capability which could be used, although I don't believe we set a password for the single VNC which is running now.
The way it works is that VNC runs restricted to localhost. The user gets a host account with the shell set to /usr/sbin/nologin. The user connects via an ssh tunnel. There is a per user sshd configuration on the host:
If I understand right, the sshd configuration allows the user to connect to only his assigned single localhost port.
I always am super glad to have your input into whatever FOSSVPS is doing and into whatever I am doing! I really appreciate your constantly generous willingness to help. Thank you so much!
Hopefully I can do much better at publishing all the details as things continue to develop. If you see the least problem with anything, I always want you to speak up just like you are doing now!
Thanks also to @FrankCastle for developing the VNC configuration!