I have heard of several cloud screw-ups as well, leading to charges of several thousands.
On one side this can happen if you experiment something outside of the free machine(s), on the other side you have all the reporting and notification tools to avoid surprises.
Nonetheless, I still see your point, reason why I prefer to use an almost dry revolut prepaid for all the cloud accounts instead of my main credit card.
As you wish, indeed the only free offer without credit cards is the one of azure for students. In any case you are not anonymous.
Considering the small audience and purpose, I would not have any problem using the always free offerings of either Oracle or Google (the latter especially if located in the US).
I don't know, wouldn't the Hypervisor be able to track resources usage by itself without anything else?
One of the answers in this discussion made me realize that this database is required only by the identity provider which I can change from the default. Considering this is a far less essential dependency than I thought I will get rid of it completely.
Thanks for the suggestion, but my 24 GB are well employed already. I wanted at least to outsource the VPN manager to a smaller VPS.
Running without docker is out of question, is a bundle of 6 docker containers. Deployment and management without it would be too complicated. Luckily somebody in another reply made me realize that the RAM eating container (cockroach DB) is far less essential than I thought and I can look for a replacement.
The one managing my VPS, controlled solely by Oracle corporation
Thanks, this is a really good point, I can try to replace the identity provider! I did not realized that cockroachDB was only a Zitadel requirement! There are many great alternatives for mesh VPNs, netmaker, nebula, and headscale as you mentioned and all of them are much lighter. I ended up hosting netbird as it is natively able to traverse my corporate NAT (maybe headscale could do it as well, I did not try it since I do not like having to configure registry keys on windows clients and losing the kernel wireguard speed on linux clients) .
Yes exactly, burstable instances it's common jargon for AWS and GCP, but applies to all major providers.
I forgot to mention, I had plenty of swap available, now I disabled swap to force zram usage. I still need to see what happens running with both, it's hard when each trial takes 12-24 hours to show its result.
In the past two weeks I set up a new VPS, and I run a small experiment. I share the results for those who are curious.
Consider that this is a backup server only, meaning that there is no outgoing traffic unless a backup is actually to be recovered, or as we will see, because of sshd.
I initially left the standard "port 22 open to the world" for 4-5 days, I then moved sshd to a different port (still open to the whole world), and finally I closed everything and turned on tailscale. You find a visualization of the resulting egress traffic in the image. Different colors are different areas of the world. Ignore the orange spikes which were my own ssh connections to set up stuff.
Main points:
there were about 10 Mb of egress per day due just to sshd answering to scanners. Not to mention the cluttering of access logs.
moving to a non standard port is reasonably sufficient to avoid traffic and log cluttering even without IP restrictions
Tailscale causes a bit of traffic, negligible of course, but continuous.
I like it, it was released a couple of days ago so something might require a bit more polishing but overall it looks better to me.