TierHive General thread | Discuss, Updates, questions & Suggestions | LATEST: new location USA WEST

13468914

Comments

  • flipsflips OG
    edited May 14

    This probably answerred before, but if I want to install something that's not on netboot.xyz or in the list, (how) can I? :D

    Thanked by (1)Not_Oles
  • @AnthonySmith said:

    @sshbox said: From this perspective a user content domain would be a welcome addition, as previously alluded to.

    It's on the list, but the list is long, it adds a surprising amount of complexity and abuse overhead to worry about, but it will happen.

    No worries, no hurries.

  • AnthonySmithAnthonySmith AdministratorHosting ProviderOGSenpai

    @flips said:
    This probably answerred before, but if I want to install something that's not on netboot.xyz or in the list, (how) can I? :D

    Create a tftpd on a 128mb box, set up your dhcp scope in the panel with a next server option pointing to your tftpd and pxelinux.0 or ipxe file or whatever you are using, and netboot the main VPS using ipxe or just use ipxe to chainload the iso or use netboot.xyz to chainload the iso it supports that, lots of options :D

    It really depends on what the something is, if they provide a cloud-init compatible image I can just add it.

    Thanked by (2)flips WSS

    TierHive - Hourly VPS - NAT Native - /24 per customer - DE, UK, SG, CA, USA x3, FR, AU, PL, NL
    FREE tokens on sign up, try before you buy. | Join us on Reddit

  • flipsflips OG
    edited May 14

    @AnthonySmith said:
    Create a tftpd on a 128mb box, set up your dhcp scope in the panel with a next server option pointing to your tftpd and pxelinux.0 or ipxe file or whatever you are using, and netboot the main VPS using ipxe or just use ipxe to chainload the iso or use netboot.xyz to chainload the iso it supports that, lots of options :D

    It really depends on what the something is, if they provide a cloud-init compatible image I can just add it.

    Ah, yes, many options ... This is all KVM based now?
    The 2 specific distros I was thinking of, was NetBSD (found a qcow2 file of v11.0-RC4, can send you URL), and the other was one where netboot.xyz failed last time I tried, either a Gentoo based one (SystemRecueCD?) or Devuan, can't recall now ... :)

    (NetBSD ISO through grub-imageboot failed, I tried on Debian KVM through two different providers, Virmach and SkylonHost)

  • WSSWSS OG

    @AnthonySmith said:

    @WSS said:

    @AnthonySmith said: It's on the list, but the list is long, it adds a surprising amount of complexity and abuse overhead to worry about, but it will happen.

    I offered you several! You want heavenfor.bid or liverwurst.party?

    I appreciate it, but I am just not ready for them yet, I dont want to take them and end up just holding on to them for months unused.

    I have a couple of stupid ones I will test with first, like pop2.me

    I'm not doing anything with them. Hell, I reregistered workgroups.win for 10 years before giving up on running my own nameservers.

    "It's a hard life- to be a stick insect." - Karl Pilkington

  • AnthonySmithAnthonySmith AdministratorHosting ProviderOGSenpai

    @flips said:

    @AnthonySmith said:
    Create a tftpd on a 128mb box, set up your dhcp scope in the panel with a next server option pointing to your tftpd and pxelinux.0 or ipxe file or whatever you are using, and netboot the main VPS using ipxe or just use ipxe to chainload the iso or use netboot.xyz to chainload the iso it supports that, lots of options :D

    It really depends on what the something is, if they provide a cloud-init compatible image I can just add it.

    Ah, yes, many options ... This is all KVM based now?
    The 2 specific distros I was thinking of, was NetBSD (found a qcow2 file of v11.0-RC4, can send you URL), and the other was one where netboot.xyz failed last time I tried, either a Gentoo based one (SystemRecueCD?) or Devuan, can't recall now ... :)

    (NetBSD ISO through grub-imageboot failed, I tried on Debian KVM through two different providers, Virmach and SkylonHost)

    well you can "bring your own image" by booting in to any rescue and using:

    wget -O - "http://yourimageurl.com/node.qcow2" | dd of=/dev/vda bs=4M 
    

    that should work, obviously you need to configure networking etc, I dont think netbsd has cloud-init support, could be wrong.

    Thanked by (1)Not_Oles

    TierHive - Hourly VPS - NAT Native - /24 per customer - DE, UK, SG, CA, USA x3, FR, AU, PL, NL
    FREE tokens on sign up, try before you buy. | Join us on Reddit

  • WSSWSS OG

    Older netboot.xyz had BSD installs. Can't remember if I ever got NetBSD to work. OpenBSD's ramdisk was all I ever needed though.

    "It's a hard life- to be a stick insect." - Karl Pilkington

  • Not_OlesNot_Oles Hosting ProviderContent Writer
    edited May 14

    Still trying, every day, to provision an instance in Kansas City, closest POP to me. :) Meanwhile, quite happy with instances in Vint Hill! <3

    Edit to add: Southern California or Arizona, some day, maybe? <3

  • WSSWSS OG

    @Not_Oles said:
    Still trying, every day, to provision an instance in Kansas City, closest POP to me. :) Meanwhile, quite happy with instances in Vint Hill! <3

    Just rent your entirely own box. They start at $9/mo.
    https://www.wholesaleinternet.net/dedicated/
    https://nocix.net/dedicated/

    "It's a hard life- to be a stick insect." - Karl Pilkington

  • Not_OlesNot_Oles Hosting ProviderContent Writer

    I must be missing something about the Recipes. . . . I tried three different recipes (Caddy, Grav, and Larvel), but none of them would let me click the final Deploy Recipe button on the recipe configuration, which stays greyed out, even as I triple check the dialog fields with red asterisks. Tried multiple locations. Tried increasing resource allocations.

    By the way, if I want to read a recipe before deploying it, how can I do that? Thanks! <3


  • AnthonySmithAnthonySmith AdministratorHosting ProviderOGSenpai

    @Not_Oles said: I must be missing something about the Recipes. . . . I tried three different recipes (Caddy, Grav, and Larvel), but none of them would let me click the final Deploy Recipe button on the recipe configuration, which stays greyed out, even as I triple check the dialog fields with red asterisks. Tried multiple locations. Tried increasing resource allocations.

    you have to click calculate costs first, this probably need improving i will make a note.

    @Not_Oles said: By the way, if I want to read a recipe before deploying it, how can I do that? Thanks!

    Also on the list.

    the recipes system has its own support/bug report system built in, if you find a problem with a recipe you can report it and the community member that published it will get a ticket, if they dont deal with the issue or respond within 14 days the recipe is removed from public view.

    Thanked by (1)Not_Oles

    TierHive - Hourly VPS - NAT Native - /24 per customer - DE, UK, SG, CA, USA x3, FR, AU, PL, NL
    FREE tokens on sign up, try before you buy. | Join us on Reddit

  • @AnthonySmith said:

    @flips said:

    @AnthonySmith said:
    Create a tftpd on a 128mb box, set up your dhcp scope in the panel with a next server option pointing to your tftpd and pxelinux.0 or ipxe file or whatever you are using, and netboot the main VPS using ipxe or just use ipxe to chainload the iso or use netboot.xyz to chainload the iso it supports that, lots of options :D

    It really depends on what the something is, if they provide a cloud-init compatible image I can just add it.

    Ah, yes, many options ... This is all KVM based now?
    The 2 specific distros I was thinking of, was NetBSD (found a qcow2 file of v11.0-RC4, can send you URL), and the other was one where netboot.xyz failed last time I tried, either a Gentoo based one (SystemRecueCD?) or Devuan, can't recall now ... :)

    (NetBSD ISO through grub-imageboot failed, I tried on Debian KVM through two different providers, Virmach and SkylonHost)

    well you can "bring your own image" by booting in to any rescue and using:

    wget -O - "http://yourimageurl.com/node.qcow2" | dd of=/dev/vda bs=4M 
    

    that should work, obviously you need to configure networking etc, I dont think netbsd has cloud-init support, could be wrong.

    If you write the image from inside the VM, you wouldn't want to write the qcow2 image (you'll want a raw image for that).

    I do have a small Lua script for NetBSD that does the cloud-init configuration (from the CIDATA CDROM). But last time I tried, the CIDATA drive disappears after the first reboot on TierHive, so the newly written image doesn't see the configuration any more (and the cloud-init network configs do use OS-specific device names)

    Thanked by (2)Not_Oles flips
  • AnthonySmithAnthonySmith AdministratorHosting ProviderOGSenpai

    @cmeerw said: If you write the image from inside the VM, you wouldn't want to write the qcow2 image (you'll want a raw image for that).

    yeah, probably need to use qemu-img convert to raw which can be done in stream also if you use a rescue system that has it

    Thanked by (1)Not_Oles

    TierHive - Hourly VPS - NAT Native - /24 per customer - DE, UK, SG, CA, USA x3, FR, AU, PL, NL
    FREE tokens on sign up, try before you buy. | Join us on Reddit

  • Not_OlesNot_Oles Hosting ProviderContent Writer

    @AnthonySmith said:
    you have to click calculate costs first,

    Completely missed calculating the costs! :star: I mean, yes, I saw the button to calculate the costs, but a cost already had been shown in the overall recipe listing (see screenshot). I guess the cost changes as the VPS resources are changed and what's shown in the overall recipe corresponds to the default resources presented inside the deploy recipe dialog.

    It's as yet unclear to me whether the prices are the same for a basic, non-recipe VPS compared to an equivalently resourced VPS deployed via a recipe.

    For the Caddy recipe, I tried both 256 and 512 MB RAM. Both times, after clicking to calculate the price, I could click the create button and the creation started . . . but both ultimately failed. Will try again before too long! Thanks! <3 Best wishes, and no rush at all. . . . :)


  • @AnthonySmith said:

    @nameserver said: Interesting wireGuard mesh networking as well. If you don't mind sharing, what software was used for virtualisation?

    yeah we had to engineer a complete custom conductor for the mesh, it probably was not worth the massive engineering effort it took to get right but it was one of those things that was always part of the design from day one so about an extra 3 - 4 weeks of time was pumped into it, happy with the result, but I would not do it again if I was starting over.

    It's all bog-standard Qemu/KVM with libvirt but we have put a huge amount of effort into optimising the nodes so squeeze every last drop of performance and density out of them. The TierHive panel itself is proprietary.

    Cool! Will it ever be a Public IPv4 Networking available? Like to monthly rent a single IPv4?

    Also I noticed a lot of location is using OVH. Which line of OVH server are you using? ECO? If you don't mind sharing.

    Thanks.

    i ran a free service -> CloudRAM.Download | Made with love | And I am a Chickenist -> Chicken.ist

  • AnthonySmithAnthonySmith AdministratorHosting ProviderOGSenpai

    @Not_Oles said: . but both ultimately failed. Will try again before too long

    please use the report button so they dont stick around if broken.

    the cost on the card is the base cose based on the recipe created suggested specs, but the deployment lets you alter that, maybe they suggest 256mb thats what its based on, but you can change it to 1GB if you want and then obviously the costs change.

    That can probably be fully automated now but the initial idea was to ensure people knew what they were about to actually spend before they spent it to remove confusion.... i guess that did not work haha.

    Thanked by (1)Not_Oles

    TierHive - Hourly VPS - NAT Native - /24 per customer - DE, UK, SG, CA, USA x3, FR, AU, PL, NL
    FREE tokens on sign up, try before you buy. | Join us on Reddit

  • AnthonySmithAnthonySmith AdministratorHosting ProviderOGSenpai

    @nameserver said:

    @AnthonySmith said:

    @nameserver said: Interesting wireGuard mesh networking as well. If you don't mind sharing, what software was used for virtualisation?

    yeah we had to engineer a complete custom conductor for the mesh, it probably was not worth the massive engineering effort it took to get right but it was one of those things that was always part of the design from day one so about an extra 3 - 4 weeks of time was pumped into it, happy with the result, but I would not do it again if I was starting over.

    It's all bog-standard Qemu/KVM with libvirt but we have put a huge amount of effort into optimising the nodes so squeeze every last drop of performance and density out of them. The TierHive panel itself is proprietary.

    Cool! Will it ever be a Public IPv4 Networking available? Like to monthly rent a single IPv4?

    Also I noticed a lot of location is using OVH. Which line of OVH server are you using? ECO? If you don't mind sharing.

    Thanks.

    Eventually we will, if all goes to plan you will be able to lease 1 or more IPv4 addresses per location and then that will essentially be your region external IP, so essentially you run your own CGNAT at that stage with the entire port range for that IP, it acts as a gateway you can 1:1 a public IP to a VPS or seperate services on the backend by port, but we wont be doing direct assignments of IPv4 to VPS.

    We also plan to support BGP/BYOIP in the future but that's probably a way off yet.

    Most of it so far has been ovh eco stuff, yes, it let us expand quickly and maintain a very low price, which is the long-term goal, to stay cheap.

    Always looking for other server options though, I have said to a few people give us your E3's your wonky servers that the ipmi des every 10 minutes on, the one that cant take a 3rd disk because it has a broken port, the one that needs a specific hack because the nic has issues etc etc, we can make it work, most hosts have them, they might as well give them to us at a discount, we are not enterprise so we can work with that.

    Also @Advin of AdvinServers did us a solid with a monster VDS in Kansas at a great price, we are willing to consider nesting nodes like this also if other hosts have beefy VPS nodes with spare cores etc.

    We have the last major feature to finish, which just failed my testing, sadly and has been sent back for a bit of rework, which is the static hosting side a bit like cloudflare pages but obviously not even remotley the same scale haha, if you have made a payment ever you get acess to the free tier for life then there is a paid tier with fairly generous limits at a pretty low price it is self replicating and has built in redundancy and failover.

    Once that is done, I can start focusing on the list of all the smaller tweaks and changes to existing stuff much faster. This page has eaten about 10 hours of time between 2 - 3 people every day for a month.

    I also need to get the " What is Tierhive " page done and include what it is not in that as we have had several "WTF" tickets recently that are starting to irritate me and I guess the messaging could be better.

    to many things all at once.

    Then in June I start ripping LES apart, haha, trying to free myself up from TierHive time to spend a month or so on big changes here.

    TierHive - Hourly VPS - NAT Native - /24 per customer - DE, UK, SG, CA, USA x3, FR, AU, PL, NL
    FREE tokens on sign up, try before you buy. | Join us on Reddit

  • Not_OlesNot_Oles Hosting ProviderContent Writer

    @cmeerw said: If you write the image from inside the VM, you wouldn't want to write the qcow2 image (you'll want a raw image for that).

    I seem to remember trying this. If I remember right, something seemed, haha, a little too big to dd onto my minimal TierHive 1 GB virtual disk:

    # Get NetBSD 11 RC3 qcow image and convert to raw
    https://lowendspirit.com/discussion/comment/231448/#Comment_231448
    https://download.cmeerw.net/netbsd/netbsd-11.0_RC3.qcow2
    chronos@penguin:~$ cd /mnt/chromeos/MyFiles/Downloads/
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ ls *qcow2
    netbsd-11.0_RC3.qcow2
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ qemu-img convert -f qcow2 \
    > netbsd-11.0_RC3.qcow2 -O raw netbsd-11.0_RC3.raw
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ ls -l netbsd-11.0_RC3*
    -rw-r--r-- 1 chronos chronos  129433600 May 10 13:12 netbsd-11.0_RC3.qcow2
    -rw-r--r-- 1 chronos chronos 1073745920 May 10 13:15 netbsd-11.0_RC3.raw
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ ls -lh netbsd-11.0_RC3*
    -rw-r--r-- 1 chronos chronos 124M May 10 13:12 netbsd-11.0_RC3.qcow2
    -rw-r--r-- 1 chronos chronos 1.1G May 10 13:15 netbsd-11.0_RC3.raw
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ 
    
    friendly:/# df .
    Filesystem           1K-blocks      Used Available Use% Mounted on
    /dev/vda                975849    169167    758601  18% /
    friendly:/# 
    

    Of course I could have increased the VPS disk size slightly. But, instead, I switched to trying to install NetBSD via netboot.xyz, at which I haven't yet really started. Maybe https://nycdn.netbsd.org/pub/NetBSD-daily/HEAD//latest/amd64/installation/misc/pxeboot_ia32.bin might work ?

    Thanks yet again @cmeerw for making the qcow2 images available! <3

  • AnthonySmithAnthonySmith AdministratorHosting ProviderOGSenpai

    I will need to look into NetBSD, we really should offer it, and 100% no offence intended to @cmeerw but we really should create and maintain it ourselves.

    TierHive - Hourly VPS - NAT Native - /24 per customer - DE, UK, SG, CA, USA x3, FR, AU, PL, NL
    FREE tokens on sign up, try before you buy. | Join us on Reddit

  • Not_OlesNot_Oles Hosting ProviderContent Writer

    @AnthonySmith said: please use the report button so they dont stick around if broken.

    Tried to use the report button. . . . It didn't seem to work. :star: On the good side, here is a screenshot of what I tried to submit. Thanks! <3

  • AnthonySmithAnthonySmith AdministratorHosting ProviderOGSenpai

    thanks, I will add that to the list.

    Thanked by (1)Not_Oles

    TierHive - Hourly VPS - NAT Native - /24 per customer - DE, UK, SG, CA, USA x3, FR, AU, PL, NL
    FREE tokens on sign up, try before you buy. | Join us on Reddit

  • @AnthonySmith said: Most of it so far has been ovh eco stuff, yes

    I am running a very similar setup, OVH ECO severs, tailscale to DHCP a 10.0.0.0/8 IP block, with a proxmox cluster to run a lot of LXC container, but just for personal selfhosting.

    I suppose all the IP address (v4 and v6) will be though tunnel because I was trying to add some ipv6 block to my eco server but found out the eco line doesn't support BYOIP or a IPv6 block.

    @AnthonySmith said: support BGP/BYOIP

    This would be a great feature that I would use, but I don't think a lot of people will be using it.

    Also maybe reseller using API?

    Thanks for the replies.

    i ran a free service -> CloudRAM.Download | Made with love | And I am a Chickenist -> Chicken.ist

  • AnthonySmithAnthonySmith AdministratorHosting ProviderOGSenpai

    @nameserver said: Thanks for the replies.

    no worries.

    Well, maybe if we get to 50 loctions it will become a good place to use those sorts of features with BYOIP over our mesh :)

    probably never going to do reseller, because we cant control expectations. you can already convert your own credit in to vouchers though, we dont care if you give them away or sell them.

    I fell out of love with Proxmox many years ago. It got so bloated and fragile, our hypervisors barely use 200mb ram when sitting idle before customers hit them. They do precisely what they were built for and nothing else :)

    We do plan to release the info on increasing efficiency and capacity on hypervisors, people would squeel at the numbers right now haha, I will probably throw it at VirtFusion first there might be some wins in there for them they can intergrate and then just release the raw HV prep scripts, but we are still changing this with every new node its pretty much 70% of the fun for me.

    While proxmox is, well, proxmox.

    TierHive - Hourly VPS - NAT Native - /24 per customer - DE, UK, SG, CA, USA x3, FR, AU, PL, NL
    FREE tokens on sign up, try before you buy. | Join us on Reddit

  • @AnthonySmith said:

    @nameserver said: Thanks for the replies.

    no worries.

    Well, maybe if we get to 50 loctions it will become a good place to use those sorts of features with BYOIP over our mesh :)

    probably never going to do reseller, because we cant control expectations. you can already convert your own credit in to vouchers though, we dont care if you give them away or sell them.

    I fell out of love with Proxmox many years ago. It got so bloated and fragile, our hypervisors barely use 200mb ram when sitting idle before customers hit them. They do precisely what they were built for and nothing else :)

    We do plan to release the info on increasing efficiency and capacity on hypervisors, people would squeel at the numbers right now haha, I will probably throw it at VirtFusion first there might be some wins in there for them they can intergrate and then just release the raw HV prep scripts, but we are still changing this with every new node its pretty much 70% of the fun for me.

    While proxmox is, well, proxmox.

    Incus is a very interesting project, supports KVM beside LXC. Very light weight, and I had fun playing with it. Have a full API as well.

    i ran a free service -> CloudRAM.Download | Made with love | And I am a Chickenist -> Chicken.ist

  • @nameserver said: I suppose all the IP address (v4 and v6) will be though tunnel because I was trying to add some ipv6 block to my eco server but found out the eco line doesn't support BYOIP or a IPv6 block.

    Does your OVH server have a IPv6 /128 IP? If so you can run this script to check if you can use the full /64 subnet:
    https://gist.github.com/loayai/8eb1fa456246202a4deece7780725da6

    Discussion on: https://lowendspirit.com/discussion/10309/kimsufi-unable-to-get-ipv6-connectivity-in-guests-either-ndppd-or-static-neighbour

    I speak fluent sarcasm and broken logic. | I would agree with you, but thæn we’d both be wrong.

  • @somik said:

    @nameserver said: I suppose all the IP address (v4 and v6) will be though tunnel because I was trying to add some ipv6 block to my eco server but found out the eco line doesn't support BYOIP or a IPv6 block.

    Does your OVH server have a IPv6 /128 IP? If so you can run this script to check if you can use the full /64 subnet:
    https://gist.github.com/loayai/8eb1fa456246202a4deece7780725da6

    Discussion on: https://lowendspirit.com/discussion/10309/kimsufi-unable-to-get-ipv6-connectivity-in-guests-either-ndppd-or-static-neighbour

    Thanks! Just configured everything in proxmox to have a DHCP working for the /64 block. radvd and ndppd was needed for this.

    i ran a free service -> CloudRAM.Download | Made with love | And I am a Chickenist -> Chicken.ist

  • @nameserver said:

    @somik said:

    @nameserver said: I suppose all the IP address (v4 and v6) will be though tunnel because I was trying to add some ipv6 block to my eco server but found out the eco line doesn't support BYOIP or a IPv6 block.

    Does your OVH server have a IPv6 /128 IP? If so you can run this script to check if you can use the full /64 subnet:
    https://gist.github.com/loayai/8eb1fa456246202a4deece7780725da6

    Discussion on: https://lowendspirit.com/discussion/10309/kimsufi-unable-to-get-ipv6-connectivity-in-guests-either-ndppd-or-static-neighbour

    Thanks! Just configured everything in proxmox to have a DHCP working for the /64 block. radvd and ndppd was needed for this.

    Congrats! OVH may have given up on this, but you made it work :lol:

    I speak fluent sarcasm and broken logic. | I would agree with you, but thæn we’d both be wrong.

  • @Not_Oles said:

    @cmeerw said: If you write the image from inside the VM, you wouldn't want to write the qcow2 image (you'll want a raw image for that).

    I seem to remember trying this. If I remember right, something seemed, haha, a little too big to dd onto my minimal TierHive 1 GB virtual disk:

    # Get NetBSD 11 RC3 qcow image and convert to raw
    https://lowendspirit.com/discussion/comment/231448/#Comment_231448
    https://download.cmeerw.net/netbsd/netbsd-11.0_RC3.qcow2
    chronos@penguin:~$ cd /mnt/chromeos/MyFiles/Downloads/
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ ls *qcow2
    netbsd-11.0_RC3.qcow2
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ qemu-img convert -f qcow2 \
    > netbsd-11.0_RC3.qcow2 -O raw netbsd-11.0_RC3.raw
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ ls -l netbsd-11.0_RC3*
    -rw-r--r-- 1 chronos chronos  129433600 May 10 13:12 netbsd-11.0_RC3.qcow2
    -rw-r--r-- 1 chronos chronos 1073745920 May 10 13:15 netbsd-11.0_RC3.raw
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ ls -lh netbsd-11.0_RC3*
    -rw-r--r-- 1 chronos chronos 124M May 10 13:12 netbsd-11.0_RC3.qcow2
    -rw-r--r-- 1 chronos chronos 1.1G May 10 13:15 netbsd-11.0_RC3.raw
    chronos@penguin:/mnt/chromeos/MyFiles/Downloads$ 
    
    friendly:/# df .
    Filesystem           1K-blocks      Used Available Use% Mounted on
    /dev/vda                975849    169167    758601  18% /
    friendly:/# 
    

    Of course I could have increased the VPS disk size slightly. But, instead, I switched to trying to install NetBSD via netboot.xyz, at which I haven't yet really started. Maybe https://nycdn.netbsd.org/pub/NetBSD-daily/HEAD//latest/amd64/installation/misc/pxeboot_ia32.bin might work ?

    Thanks yet again @cmeerw for making the qcow2 images available! <3

    The images currently have a fixed-size 512 MB swap partition (which I think is fine for somewhat larger VPSes) - maybe I should get rid of that or create it on first boot. The NetBSD root partition is actually just 512 MB.

    Thanked by (2)Not_Oles flips
  • @AnthonySmith have you seen how the tierhive logo (your forum thumbnail) looks in the dark theme?

    Cause I for sure cant see it :lol:

    Thanked by (1)AnthonySmith

    I speak fluent sarcasm and broken logic. | I would agree with you, but thæn we’d both be wrong.

  • @AnthonySmith said:
    I will need to look into NetBSD, we really should offer it, and 100% no offence intended to @cmeerw but we really should create and maintain it ourselves.

    Feel free to start with the scripts I have on https://bitbucket.org/cmeerw/netbsd-cloud-image/ (btw, I am also creating raw images with smaller swap here, but maybe I should do something like dynamically creating swap on first boot)

    There are actually a few things that are a bit tricky for NetBSD: you want to end up with a file system that has been resized to use all available space, but also is a FFSv2 filesystem and has wapbl enabled (otherwise performance suffers) - the tricky part here is that once you have wapbl enabled, you aren't supposed to resize the filesystem any more.

    The other thing is that the full cloud-init is a monster and pulls in a lot of dependencies, so a full cloud-init doesn't really fit with TierHive. Having a BSD image that takes 5 GB (OpenBSD) or 10 GB (FreeBSD) doesn't make sense to me.

    Thanked by (3)WSS flips Not_Oles
Sign In or Register to comment.