Cloudron Forum

@nebulon said in Managing SSL certs via Cloudron CLI: you have to "forget" the page in your browser yes, or visit the site in an incognito session. Clearing these entries from the profile in Chrome is slightly more complicated, but doable as well. https://msutexas.edu/library/clearhsts.php

@mastadamus If you use namecheap API, you don't need port 80. This is because Cloudron will use Let's encrypt DNS automation to get certs. Note that this will require you to sometimes type "https://" explicitly in some browsers because some browsers will default to connecting on port 80 and then the redirect will take it to the https site. In addition, Cloudron has HSTS, so future connects will directly be to 443 and no redirect dance.

@eivlil01 said in Using my own certificates: @girish I'm using the DNS based challenge, but for a wildcard entry. Cloudron only supports http based challenge for wildcards, or DNS based challenge but then it creates one entry per app. Ah, I see what you are saying now. So you have a wildcard DNS entry pointing to the server but also use DNS automation to get wildcard certs. Indeed, Cloudron does not support that.

@girish so I created a new WP install on a different cloudron for the domain, https://slappersonly.co -- everything seems in order now, even for people who had errors before. Meanwhile I switched the older WP install to a new domain on the original cloudron https://slaps.vip .. there do not seem to be any issues for either domain now. Not too terribly inconvenient as the 2 sites serve different purposes for the same brand, but bizarre nonetheless.

@jdaviescoates nope, just regular domains, not sure why but will just wait and see I guess

@niko You have to convert the app into a Cloudron app for all this to reliably work. We don't support running/installing other things other than Cloudron on the same server. This is because Cloudron will overwrite nginx configuration etc from time to time (for example, updates bring in new configuration). If your app has a Dockerfile, you can make it a custom app with not too much work - https://docs.cloudron.io/custom-apps/tutorial/ . Custom app will automatically get certs, backups, restore, clone features etc with no extra work. What framework/language does your app use?

Since, we got so many support tickets about this already I will paste what I said in the other thread. Let's Encrypt have started using R3 as the intermediary cert - https://scotthelme.co.uk/lets-encrypts-new-root-and-intermediate-certificates/ . This cert has issuer text slightly different. Since the text has changed, Cloudron tries to renew the certs too early and this results in the above notification. The notification can be ignored since it's a false alarm, the certs and sites will be fine. There are two ways to fix this: Update to Cloudron 6 - you can go to Settings -> Check For Updates and then Update. It will give a notification that it is unstable. It's reasonably safe to update, the notification exists because we roll out updates very slowly to keep support manageable for us. Please expect some downtime (like 10 mins) since the update re-configures all the docker containers. Alternately, you can make this one line change in your current Cloudron version - https://git.cloudron.io/cloudron/box/-/commit/3e62f1913ab05750a343c197c519d38bf17d5b3b . The file is /home/yellowtent/box/src/reverseproxy.js and then systemctl restart box.

@girish feeling lazy, will wait for the official update

So I'm pretty convinced the issue was the way I wrote the CAA records. I think my DNS provider didn't need the double-quotes in there and it caused issues. Reason I say that is because after introducing the CAA records, I suddenly had the certificate renewal errors. Then when using a DNS check tool and I looked up CAA records for Google and Mozilla and more, none of them had the double-quote in there, but mine did. So I am sure that was the issue, as everything worked fine again after I removed the double-quotes. I suspect the double-quotes was being taken literally as a string and so letsencrypt.org is not the same as "letsencrypt.org" in the DNS CAA record. I was able to later find the logs I had seen in the early morning which shows the following which confirms my conclusion: CAA record for <domain> prevents issuance. So for anyone who comes across this later, make sure you're not using double-quotes I guess. haha.

@wu-lee do you know why it had failed to renew previously?

@d19dotca Ah, I see this on our server as well. It seems that this is printed for all incoming mails because Haraka tries to transfer mail to Dovecot via TLS (via LMTP). Disabling SSL removes the message. Since the message transfer is internal, it's fine. I have pushed the fix for next release but the message is safe to ignore.

@girish Sorry for the late reply, I the default forum settings were to send me email notifications once a week, I have changed that to daily. I was trying to use the self-signed certificate for all apps. I have gone ahead and followed the steps in the blog article I linked to and was able to upload the full CA and web server cert chain to Cloudron and after getting the root certificate authority into all the devices accessing Cloudron everything is working well from Windows and iOS based devices. It looks like Apple is going to be decreasing the certificate lifetime further down to 398 days so you may want to lower the lifetime down to a year to get ahead of that change.

@andrewj720 Looks like DNS is not working on your server. You can also try host cloudron.io etc, I guess none of it working? Can you check if your cloud firewall allows outbound port 53 UDP ? I think there was a post on this forum some time ago that someone had it blocked in AWS security group by mistake, for example.

This has helped me in version Cloudron v7.3.2

@marcusquinn said in Lets Encrypt renewal time: Without looking at that screen again, maybe it wasn't clear it should recommend using the root domain for that input? I think many people start out just like you did and then move it to the main domain. We don't put the recommendation as such because I think it can be scary to throw your root domain and API credentials into a product you are just first trying out.

@marcusquinn said in Is there a way to set the LetsEncrypt email separately?: So Superadmin's are Owners then? In that case I have about 20 Indeed! You can downgrade everyone to be an admin. The main difference between superadmin and admin are that superadmins is meant to be the person who has access to the server (and the one who set things up initially). Superadmin also manages the subscription and has acess to mail server logs. Admins don't have access to these two things. Ideally, there is only one superadmin. We wanted to enforce this but migration from previous setups proved to be a bit problematic.

@nebulon Like I said, I thought it would be an issue for a particular client, but it's not! However, only as long as the config stays as it is; if you ever strengthen it by adding the "includeSubdomains" directive in the HSTS header (as it is advised sometimes in some of the readings I found for better security), you could cut access to subdomains that are not managed by Cloudron and cannot do TLS. The typical fictional scenario, if the config ever changes, would be: www.watering-plants-automatically.cloud is a website from a company that offers to manage clients' gardens; the designer of the website hosts it on its Cloudron instance for the client. The company already has stuff they host on subdomains, and won't relinquish access to the DNS server for security reasons. However www and the root domain both point to the Cloudron server IP, so Let's Encrypt works fine in "Manual" mode in the Domains & Certs tab of the designer's Cloudron; An end-user visits the website, decides to sign up and pay for their new fangled tool, which is hosted at myplants.watering-plants-automatically.cloud. This subdomain points to the IP of the appliance that manages the users' gardens. This is an old, crummy box that won't allow TLS, because it's almost an antic at this point; The user cannot connect to their tool, and throws an HSTS error. It's not an issue yet, but it might be something to think about if you ever consider changing the configuration (let's say, if you decide all domains with a wildcard cert should have includeSubdomains in their HSTS headers). Security-wise, it makes a ton of sense: let's say you type http://www.domain.tld in your browser. The server 302s you to https://www.domain.tld which has the HSTS header and "includeSubdomains" You later type http://mail.domain.tld in your browser: the browser will immediately connect to https instead, avoiding potential MITM attacks. Pretty powerful, but it might be an issue in this particular case where some subdomains shouldn't be covered. I initially thought I read in the docs that the HSTS config was such that all subdomains were included, and I remember that before using Cloudron, for this specific client, I set the header to "includeSubdomains", which promptly disallowed access to many tools I do not host because they didn't support TLS on them, if the user visited the main website before. So yeah, feel free to close that topic, because it's not an issue unless you decide to change the config server-wide

@girish Yes I did, and the problem with the certificates is now fixed. Thank you!

@murgero Thank you bro! I learned how to create and manage SSH keys via terminal today. I'm going to start learning how to do basic Linux Terminal Commands now. So when installing Cloudron, I should select Let's Encrypt - Wildcard and turn that to yes. Anything else that you recommend?