TierHive - Update post - 15th October 2026 - Annoying first reboot issue is fixed + NetBSD

imok

@AnthonySmith said:

@imok said:
I want to move to Vancouver

In general or just a VPS move?

Me, my wife, my cats and my servers.

zed

@WSS said:

@AnthonySmith said:

It is literally a direct slap in the face to be doing this sort of thing on an ultra-budget, small-scale project on a shared IP, so zero tolerance will be the reaction to zero respect.

Your rentals, your rules. I'd start by nerfing anyone with a non-super-small Linux instance that refuses to run qemu-ga. You've basically undercut VirMach as the cheapest host plausible with actual service provided. You're going to attract the MJJs.

i'll accept my refund i guess. shame.

edit: support zero tolerance for bullshit on a shared ip, regardless.

WSS

@zed said: i'll accept my refund i guess. shame.

https://www.youtube.com/shorts/r7FPyARszHc

zed

@WSS said:

@zed said: i'll accept my refund i guess. shame.

https://www.youtube.com/shorts/r7FPyARszHc

i dont youtube sorry sure it's on point tho!

WSS

For whatever reason, the backup is still exactly the same size. Yes, I did a complete shutdown.

AnthonySmith

The status page is finally done, it is currently only available to logged in users but over the week it will be pushed in to the public site.

Announcements now go here too instead of filling up the dashboard, timestamped and kept forever as requested. Downtime incidents will link to the related announcement too if one exists going forward.

There is also a ping test and a network map page that shows each location's status and mesh links in fairly real time.

You can also check the latency between locations from here and zoom in to a regional level:

And run ping tests from 1 or all locations to the IP of your choice (rate-limited to prevent flooding or abuse)

yoshiki

@AnthonySmith Do you see a South America (Chile, Brazil) point anytime soon, or in the long-run even?

AnthonySmith

@yoshiki said:
@AnthonySmith Do you see a South America (Chile, Brazil) point anytime soon, or in the long-run even?

I would love to do South America, but whenever I ask around, there is very little feedback on hosts that have suitable hardware in those regions we could use.

We will do Miami fairly soon, though maybe in a few weeks, which may be of some use to users in South America.

aphs

Is there a IP to ping for each location (can be the .1 not the actual NAT)? Trying to figure out which aren't OVH

AnthonySmith

@aphs said:
Is there a IP to ping for each location (can be the .1 not the actual NAT)? Trying to figure out which aren't OVH

Kansas, New York, Japan, Vancouver are not OVH, London is mixed but very likely you wouldn't go on OVH right now, Paris, and Netherlands also not OVH.

yoursunny

@aphs said:
Is there a IP to ping for each location

Go to one of your servers and run:

tcpdump -pni eth0 icmp

Then run a ping test from one of the TierHive locations.
You can then see the TierHive IP address.

risturiz

@AnthonySmith said: We will do Miami fairly soon, though maybe in a few weeks, which may be of some use to users in South America.

Canceling my few MIA VPS :-)

davide

The server location handles used by VPS Price Tracker broke today, so I passed by TierHive to have a look and... :'(

BTW those are about ~15 locations, if each has one dedicated server, it is quite a spend, but I remember you saying it recently turned profitable, congrats on that.

sumo

@AnthonySmith said:

A bunch of accounts making tons of transactions from different cards and then filtering the tokens to a central account that has never made a single payment, so that was fun processing around 50 refunds and filing about 30 different fraudrecord reports on about 50 different IP's and emails.

That's definitely stress testing. Hopefully it didn't involve too much time or money.

@AnthonySmith said:

but I think I will limit token generation to 30+ days as that's the danger zone for chargebacks. See how it goes and see what next month brings.

Not sure if all the tokens are fungible. But, maybe only allow accounts to accept non public/admin tokens 30+ days after their first paid account balance? Or if they're not fungible, limit how much those accounts (never made a paid or made first payment less than 30 days ago) can receive in tokens, in cumulative, say 1 token?

AnthonySmith

We finally solved that annoying issue of the first reboot after deployment not actually rebooting but shutting down.

NetBSD Cloud init Template proof of concept is done, just working on the auto disk resize on provisioning and that will be released.

AnthonySmith

NetBSD is now available

Template runs with 128mb ram and 1GB disk:

pretty light weight:

london-netbsd-01# df -h
Filesystem     Size   Used  Avail %Cap Mounted on
/dev/ld0a      992M   411M   531M  44% /
kernfs         1.0K   1.0K     0B 100% /kern
ptyfs          1.0K   1.0K     0B 100% /dev/pts
procfs         4.0K   4.0K     0B 100% /proc
tmpfs           32M   4.0K    32M   1% /var/shm


load averages:  0.01,  0.02,  0.00;               up 0+00:03:31                                                                                                                                                                    16:19:56
11 processes: 10 sleeping, 1 on CPU
CPU states:  0.0% user,  0.0% nice,  0.3% system,  0.0% interrupt, 99.7% idle
Memory: 31M Act, 4980K Inact, 8652K Exec, 16M File, 21M Free
Swap: / Pools: 20M Used / Network: 1K In, 1K Out

  PID USERNAME PRI NICE   SIZE   RES STATE       TIME   WCPU    CPU COMMAND
  456 root      43    0    12M 2052K CPU         0:00  0.00%  0.00% top
    0 root     126    0     0K   16M pgdaemon    0:00  0.00%  0.00% [system]
 1122 root      85    0    24M 7804K poll        0:00  0.00%  0.00% sshd-session
 2426 root      85    0    24M 5568K poll        0:00  0.00%  0.00% sshd-session
  296 root      85    0    19M 3716K poll        0:00  0.00%  0.00% sshd
  448 root      85    0    12M 2108K wait        0:00  0.00%  0.00% sh
 2233 root      85    0    12M 1548K ttyraw      0:00  0.00%  0.00% getty
 2044 root      85    0    14M 1488K nanoslp     0:00  0.00%  0.00% cron
  734 root      85    0    20M 1468K kqueue      0:00  0.00%  0.00% syslogd
    1 root      85    0    12M 1388K wait        0:00  0.00%  0.00% init
 1268 root      85    0    12M  528K kqueue      0:00  0.00%  0.00% powerd

File system auto-resizes on first boot, which requires a reboot if you select anything over 1GB disk.

Cloud-init is installed, which means Python is also installed, so you can probably reclaim around 200MB if you want on the ultra light. I also stripped out the HW Firmware and shit it never needs on a VPS, like Wi-Fi, etc.

I am not terribly experienced with BSD to be fair so any feedback is welcome.

Enjoy!

cmeerw

@AnthonySmith said: NetBSD is now available

Thanks, looking pretty good.

Except that it doesn't like IPv6:

AnthonySmith

@cmeerw said:

@AnthonySmith said: NetBSD is now available

Thanks, looking pretty good.

Except that it doesn't like IPv6:

Looks more like a yaml issue?

cmeerw

@AnthonySmith said:

@cmeerw said:

@AnthonySmith said: NetBSD is now available

Thanks, looking pretty good.

Except that it doesn't like IPv6:

Looks more like a yaml issue?

Well... yes, the yaml that the TierHive panel generates for the NetBSD IPv6 network config is probably missing some quotes

AnthonySmith

@cmeerw said:

@AnthonySmith said:

@cmeerw said:

@AnthonySmith said: NetBSD is now available

Thanks, looking pretty good.

Except that it doesn't like IPv6:

Looks more like a yaml issue?

Well... yes, the yaml that the TierHive panel generates for the NetBSD IPv6 network config is probably missing some quotes

We don't auto-configure IPv6 at all, and absolutely do not include IPv6 addresses in the YAML, so I am very confused.

Our network data looks like this; nothing in Tierhive will attempt to auto-configure your IPv6 for you.

#network
version: 2
ethernets:
    $nic_name:
          match:
              macaddress: '$mac_address'
          addresses:
              - $static_ip/$cidr_subnet
          routes:
              - to: 0.0.0.0/0
                via: $gateway
          nameservers:
              addresses: [$dns_servers]

AnthonySmith

nevermind... I see this issue its our side, brilliant, I assume you reinstalled an existing VPS, I will sort that out.

cmeerw

@AnthonySmith said:
nevermind... I see this issue its our side, brilliant, I assume you reinstalled an existing VPS, I will sort that out.

Yes, so what I did was: create a VPS with 1 GB disk, no IPv6 and NetBSD 11. This worked fine, but I noticed I also wanted to see how the disk resizing worked and if IPv6 worked. So I then upgraded to a 2 GB disk, assigned an IPv6 address and reinstalled NetBSD 11. VPS booted, resized the disk, rebooted and then failed to auto-configure because of the yaml issue.

Not_Oles

@AnthonySmith said: I see this issue its our side, brilliant

Hi!

Thanks for adding NetBSD! That's great!

Can you please say a little more about the issue so I can try to understand?

Thanks again!

Tom

AnthonySmith

@cmeerw said:

@AnthonySmith said:
nevermind... I see this issue its our side, brilliant, I assume you reinstalled an existing VPS, I will sort that out.

Yes, so what I did was: create a VPS with 1 GB disk, no IPv6 and NetBSD 11. This worked fine, but I noticed I also wanted to see how the disk resizing worked and if IPv6 worked. So I then upgraded to a 2 GB disk, assigned an IPv6 address and reinstalled NetBSD 11. VPS booted, resized the disk, rebooted and then failed to auto-configure because of the yaml issue.

Yeah, some early IPv6 auto-config code has been pushed in that should not exist, and it's literally using my from-memory, handwritten YAML from scratch that I used to give an example. Human error. I am on Zulip now having a chat; we will get this done as an emergency fix in about 5 minutes, its a clean block to remove, apparently.

I guess the cat's out of the bag now that auto IPv6 config is coming soon.

Edit: Fixed

AnthonySmith

@cmeerw said:

@AnthonySmith said:
nevermind... I see this issue its our side, brilliant, I assume you reinstalled an existing VPS, I will sort that out.

Yes, so what I did was: create a VPS with 1 GB disk, no IPv6 and NetBSD 11. This worked fine, but I noticed I also wanted to see how the disk resizing worked and if IPv6 worked. So I then upgraded to a 2 GB disk, assigned an IPv6 address and reinstalled NetBSD 11. VPS booted, resized the disk, rebooted and then failed to auto-configure because of the yaml issue.

Should be sorted now; you might want to redeploy from scratch. (Unless you already worked around it)

WSS

@cmeerw said: Except that it doesn't like IPv6:

I was going to tease @AnthonySmith that even less people use NetBSD than IPv6, but the joke wrote itself when I was out working.

AnthonySmith

@WSS said:

@cmeerw said: Except that it doesn't like IPv6:

I was going to tease @AnthonySmith that even less people use NetBSD than IPv6, but the joke wrote itself when I was out working.

I thought of you immediately....

WSS

@AnthonySmith said: I thought of you immediately....