• 0 Posts
  • 15 Comments
Joined 1 year ago
cake
Cake day: July 30th, 2023

help-circle





  • So the opening of ports works, but it’s not the most secure or best way to do it imo… what happens is the certbot registers with letsencrypts api and attempts to request a cert via http challenge, it then hosts a small website with a code from letsencrypt to prove that you do in fact own the domain and are who you say you are. Let’s encrypt then goes to the url, verifies it sees the text, and issues a cert to the certbot. The problem here is you have to open these ports to the internet, and they need to be open when certificates are renewed (let’s encrypt only issues a 90 day cert)… if you want to leave those ports open that’s not exactly a safe practice, and manually doing it every 3 months is less than ideal…

    With dns challenge, the certbot uses the api of your dns provider (cloudflare or porkbun), the process is similar, it talks with letsencrypt, let’s encrypt gives it a string and a dns record it expects to see, then certbot talks to your dns, makes a txt record with the string provided, then let’s encrypt checks for that dns record, if it finds it, it issues a cert to the certbot. In this scenario, certbot is connecting out to your dns provider and making the record for you, no opening of ports. And if you leave the api key active, it will auto renew on a schedule so you don’t have to really worry about it.

    I highly recommend looking into dns challenge some more, watch some videos on it there are lots on YouTube.

    As for the dns record, not sure if it’s not allowing the wildcard record or what but I wouldn’t use *.example.com, make an entry for the actual host/service you are hosting, like portainer.example.com.



  • I like the “1 liter PCs/home servers” for this kind of stuff. I have a 3 node proxmox cluster running on hp elitebook mini 800 g4’s. I got them for around $120 each on ebay (prices vary). Other big manufacturers have their own mini modules (hp, lenovo, dell) Generally these have a lower price tag than something like a similar generation intel nuc because it’s less of a niche market, these are used in business office environments and usually sold used pretty cheap when hardware is refreshed or businesses are closed. You can find replacement parts easily also. Just make sure they include a power adapter if you do search for one.

    Mine are running i5-8500t processor which supports Intel quicksync and performs very well for video transcoding in plex. Should easily be able to do a couple of 4k transcodes easily. If you’re not interested in running proxmox, this would run OMV easily and have plenty of power to run lots of containers.


  • I use nginx proxy manager with dns challenge to get a *.example.com cert that I then use to host services internally. I just checked, it supports dns challenge for porkbun, you may want to give it a try again. Also, you shouldn’t really need to forward dns to duckdns. You can have public dns records point to an internal ip.

    This is what I do, I have example.com (dns registered with cloudflare but should work the same with porkbun) I then create an a record for portainer.example.com to 192.168.0.5.

    Internally my nginxproxymanager is running at 192.168.0.5 and portainer is running at https://192.168.0.6:9443

    Then in nginxproxymanager I create a dns challenge (you’ll have to look up some videos on how to do this, it’s not very difficult it usually just takes a api key and secret key) then I create a new proxy host for portainer.example.com pointing to https://192.168.0.6:9443 and you select the *.example.com as your ssl cert for the proxy host

    Now internally go to https://portainer.example.com and it should work.







  • Well with bitwarden/vaultwarden you can have a copy of your entire vault on your phone or computer or both… so even if your server was totally dead, you’d have access to your passwords. Solid backups is a must, I follow the 3-2-1 rule on super critical systems (like vaultwarden) and test that you can actually recover. Something as simple as spinning up a VPS, testing a restore, testing access, see if that could work in a pinch until you get your server back online, then tear it down. Linode is very cheap for this kind of testing, it’d only cost you a few pennies to run a “dr” test of your critical systems. Of course you still want to secure it, I’d recommend wireguard or tailscale instead of opening access to your DR node to the internet, but as a temporary test it’s probably fine if your running patched up to date versions of docker, vaultwarden, and I’d always recommend putting a reverse proxy in front like nginx.