Any fellow Hetzner users here ? I’ve been using Hetzner for the VPS that this platform runs on for some time now. Performance-wise, it’s perfect. I have a 4 x CPU, 160Gb SSD disk, and 16Gb RAM server running and it’s pretty stable - apart from the server suddenly going offline every x days due to a bug in the IPv6 DHCP package. The workaround for this was to set a static IP instead - see this
https://docs.hetzner.com/de/cloud/servers/static-configuration/
Essentially, it means I had to configure this server as below
Existing netplan
# This file is generated from information provided by the datasource. Changes
# to it will not persist across an instance reboot. To disable cloud-init's
# network configuration capabilities, write a file
# /etc/cloud/cloud.cfg.d/99-disable-network-config.cfg with the following:
# network: {config: disabled}
network:
version: 2
ethernets:
eth0:
dhcp4: true
addresses: ['ipv6 address']
gateway6: fe80::1
nameservers:
addresses: [127.0.0.53, 127.0.0.1]
match:
macaddress: macaddress
set-name: eth0
Proposed change
network:
version: 2
renderer: networkd
ethernets:
eth0:
addresses:
- ip address/32
- ipv6 address/64
routes:
- to: 0.0.0.0/0
via: gateway
on-link: true
gateway6: fe80::1
match:
macaddress: macaddress
set-name: eth0
Then create file /etc/cloud/cloud.cfg.d/99-disable-network-config.cfg
Add the below
network:
config: disabled
Save, and then reboot (or at the very least, stop and restart netplan)
The point here is that those unfamiliar with the inner workings of Linux will have absolutely NO CLUE what to modify here. This is one of the downsides of having an unmanaged VPS - there is ZERO support (I’m going to be writing an independant blog article about this soon). For me, this isn’t the end of the world, but it could well mean disaster for anyone else just starting out. Then, there’s the total lack of support when I asked them about unexpectedly high TTFB (Time To First Byte), which is around 200ms for a flat HTML file - it should ideally be less than 50. In addition, this platform hits anywhere between 400-600ms for TTFB. This isn’t great, but there isn’t much I can do to reduce it (there’s another article I’m going to write about that as well). Hetzner’s response ?
First, they asked me to use mtr to conduct 1000 traces from my PC to the VPS and vice-versa. Not only does each run take around 18 minutes to complete, but you have to copy the results, and email them back to the support desk. Once these were “reviewed” (I’m using quotes here as there really isn’t any useful information as such), they reverted with
Dear Client,
the MTR does not show any issue, so we can’t see any network issue on our side. Also we’ve checked the hostsystem and can’t see any issue with it as well. Please check your server logs.
No sh*t, Sherlock. I could have told you that myself. Check my server logs ? You don’t think I’ve already done that ?
And so here’s the lesson. Hetzner absolutely EXCEL when it comes to a stable machine, cost to run, backups, and everything else - until you need support, and that’s where it all falls apart sadly.
You’re on your own there… Be warned 
