I have noticed really bad inbound latency on the VPS I host this site
on. Way slower than what skhron says it
should be on their site so clearly something was up. I noticed when pinging
example.com the first ping took a while
but after that it was fine compared to pinging its IP address which was
pretty much instant so clearly the DNS server was shit. While doing this I
had the VPS dashboard open and saw DNS server settings which I assumed
(assuming things often lead to bad outcomes) would change something on the
DHCP server. After that I noticed the network on the VPS go completely
down. Only way I could get in was with VNC and from the server I couldn't
ping anything not even IP addresses. After changing back to the default DNS
the issue continued. Turned out there is no DHCP the VPS server just
configures everything statically with some scripts and it messed up. I use
freebsd which wasn't well tested by whoever made that system so it nuked
itself.
Even after turning on manual mode and setting up the networking by
hand it was still fucked. The only thing I backed up was my git repos. All
the configs (some of them a pain in the ass to make) were stuck on there
without any way to get them off except painfully hand copy with the 80
column VNC display. Linux live environments don't support zfs and the
drivers to make that happen require rebooting which live environments reset
on boot. Uploading freebsd live images to the VPS would be painful due to
the stuppa way it handles them.
Because I didn't want to go through any of those things I contacted tech
support (I hate contacting those types of places). Most places have an
under powered LLM running the show, people who have good writing/talking
skills but have computer skills limited to using word processors and web
browsers, over worked intern who has million other things on their
mind... Skhron turned out to be one of the rare gems that has tech support
that not only knows what they are doing but also has the time/will to do
so. The tech support person actually took the time to recreate the issue on
another freebsd instance while most tech support people by now would have
sent a copy pasted corporate answer that translates to "go fuck
yourself". After a few hours they sent a manual route config which fixed it
and they also said to configure DNS from /etc/resolv.conf
instead.
Less so when using more popular operating systems on a VPS but do be careful with dashboard settings that might be changing things in the operating system itself instead of on the network and VM. If you can preform a task from a config file that is often the safer choice because you don't know what the fuck the VPS dashboard is doing behind the scenes. Also anything you don't want to loose make backups of because even if you do everything right things will still fuck up! If this issue wasn't fixed a fresh install would have been the only other option. If up time is important for you then automatic backups and full backups are a must have. Regardless keep an offline backup of important things on your own hard drive.