I also tend to find myself replacing selfhosted services with public ones. And I have realized that this is an antipattern for me because i realized that there is always a trade-off when you use a public service instead of a selfhostted one. This has made me motivated to selfhost alot more and pay the time and money to do so because if i dont pay the selfhosting tax which is predictable i will get unpredictable results from public services and selfhosting well is the most proper way anyway

Throgh please write in more detail (if possible) what board did you use to install ?

Maybe something in this regard https://www.okdo.com/getting-started/ge … ck-5b-sbc/ (when there is time) Is it 32 or 64 bit architecture?

Some useful links:

https://www.fsf.org/resources/hw/single-board-computers

https://wiki.parabola.nu/Computers

https://www.coreboot.org/Supported_Chipsets_and_Devices

it would help us very much, if there are people helping and support us in building an own keyserver.
We would like to use onak being already part in our repositories to come for release 0.4.3.1.

What do you think about Biboumi? https://biboumi.louiz.org/

However, yesterday, I wanted to help test HyperbolaBSD components and also see if I could port Anarch to HyperbolaBSD, so I git-cloned a bunch of HyperbolaBSD components from Hyperbola's git. Then, I realized that I had accidentally downloaded over a gigabyte of stuff from Hyperbola's git, so I felt quite guilty, as if I was ruining Hyperbola and driving it to bankruptcy by placing such a massive load on its servers. And I haven't yet figured out how to compile and/or run most of the stuff I have downloaded, so I was afraid that I wouldn't be able to "give back" by porting software or sending in bug reports or whatever.

Now, I'm aware that there are mirrors that help Hyperbola deal with this load, but still. What I want to ask is:
Do my attempts to be frugal server-wise help Hyperbola, and should other people adopt this approach too?
And should I continue to religiously avoid downloading any large packages in order to not place too much load upon Hyperbola, or am I misdirecting my efforts?
And should I continue to feel guilty about git-cloning over a gigabyte of stuff which I haven't yet figured out how to compile and/or run? Have I negatively affected Hyperbola and the free software movement by doing this?

I posted the following on a different forum, but I have tried it with Hyperbola and it works just as well.

I work mostly in the console, so I usually run my server VMs with serial console output only. More recently I have begun to fully encrypt the whole os, without a separate /boot partition. All went well until I realized that the cryptomount command which is called by Grub, is executed before the serial console initialization and as a result the prompt to enter the passphrase was not being passed to the serial console. I started working to fix the problem only to find out that as far as I could see, there were no simple clear examples of how to solve this specific problem. So here I am trying to detail how I fixed the problem in case this is useful to anyone else.

Cheers, and please feel free to correct/improve my post.

NOTE:

* My setup uses BIOS/MBR, but should also work for UEFI with small tweaks.
* I assume that the reader already knows how to configure grub to display in the serial console.
* Check your grub.cfg carefully for all the modules that need to be preloaded in the core.img ... i.e. luks2 vs luks
* A subsequent grub-install will blow out this configuration and it will no longer work.

I worked out the solution based on information provided by these two links:

https://wiki.archlinux.org/title/GRUB/Tips_and_tricks
https://cryptsetup-team.pages.debian.ne

    ----------------- early-grub.cfg ----------------------------------
    serial --unit=0 --speed=9600
    terminal_output serial
    terminal_input serial

    cryptomount -u YOUR_UUID

    set root=(cryptouuid/YOUR_UUID)
    set prefix=/boot/grub
    configfile grub.cfg
    ----------------------------------------------------------------------------

grub-mkimage -c early-grub.cfg  -o /boot/grub/i386-pc/core.img -O i386-pc biosdisk cryptodisk ext2 gcry_rijndael gcry_sha256 gzio luks part_gpt part_msdos serial terminal configfile keylayouts at_keyboard 
grub-bios-setup -d /boot/grub/i386-pc/ /dev/YOUR_DISK

# doas ip -6 addr add 2001:41d0:602:b80:0001:0001:133c:e8dc/64 dev ens3

# ip -6 addr show

2: ens3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 state UP qlen 1000
    inet6 2001:41d0:602:b80:1:1:133c:e8dc/64 scope global
       valid_lft forever preferred_lft forever

# ip -6 route
2001:41d0:602:b80::/64 dev ens3 proto kernel metric 256 pref medium

# From another host
# ping -6 2001:41d0:602:b80:0001:0001:133c:e8dc
PING 2001:41d0:602:b80:0001:0001:133c:e8dc(2001:41d0:602:b80:1:1:133c:e8dc) 56 data bytes

--- 2001:41d0:602:b80:0001:0001:133c:e8dc ping statistics ---
4 packets transmitted, 0 received, 100% packet loss, time 3095ms

can someone help me thanks.

please recommend me some privacy and freedom respecting CDNs, preferably free.

I have to convert ipv4 requests to ipv6 and use them as a proxy server. Will the encryption is https be from end to end (client to my server) or do I need to trust CDNs?

response.set_cookie('cookie_key', secure=True, httponly=True, samesite='Strict', max_age=1800)

Above code in python-flask ensures that the cookie is sent over a secure https connection only and prevents cross scripting attacks to some extent. The issue I get while developing is iceweasel-uxp shows 127.0.0.1:5000 is untrustworthy since it isn't encrypted for development testing.

response.set_cookie('cookie_key', httponly=True, samesite='Strict', max_age=1800)

Above code works well. My question is
Is there a development mode in iceweasel-uxp OR how do I make 127.0.0.1:5000 a trusted address?

I'm a total beginner to this sort of thing but I appreciate it needs to have
encrypted storage for my emails and be secure against possible malware and hacking attacks, which means I should rely on libre software to build the server.

So I wonder what the Hyperbola community recommends for building
an email server, besides installing it in libre hardware like Technoethical's computers?

Thanks.

I want to give some insights running Hyperbola on the system named within the subject. First some more information: It is a single-board computer with pre-installed Coreboot as base. I'm using the version with 4 GB RAM so far and installed Hyperbola on a SSD mSata I've added later. Running perfectly for CalDAV, Mailing and some webservices. Would be therefore also interesting giving the scripts back I've written for Dovecot and Radicale so both can be run without problems on OpenRC. I'll keep you updated within this thread here but after even three months I can say: Hyperbola is running without problems as serversystem and therefore I'd like to prefer it for next year upgrade-cycle - when support for Ubuntu 14.04 is running out.