Let's Encypt support
-
If you dont know your certs have been nicked, then they are useful to hackers for a whole year.
Believe me, nobody wants my certs ;D
-
Nobody realy knows if the CA you use is trustfull or has a connection to the government or an other party.
Best practice is to generate the private key you wish to use for a certificate locally on an uncompromised machine with a cryptographically secure random number generator. In this scenario the security of the private key is entirely in your hands because the Certificate Signing Request sent to the CA does not contain the private key.
-
I see no point to support of this on pfsense to be honest.. Other than maybe captive portal. For the webgui to admin pfsense.. Just use CA on pfsense to create a cert and trust it on your machines you will admin pfsense from. This really should be a really small list of machines!!
Now using https in your captive portal, yes https could be handy to be able to use a cert that is auto trusted by your guests.. But prob going to fail anyway even if the https your serving is valid for your https captive portal page since redirect of https is tricky and browsers don't like it and balk as well they should and if the site they have been going to via https is using hsts most browsers going to prevent it.
Your best solution is to just notify users that they have to hit your captive portal directly or try and go to a http site to get redirected. You could then redirect them to a https url that you have trusted cert for.. This would really be the only reason I could see for this.. But why not just get a https for a year from free like start or there are other places they cost a whole $10 a year.
If your so cheap to not want to pay for your ssl.. Just don't bother doing your captive portal via https - or use self signed and deploy that to your users of your captive portal on the your going to have anyway on the http url for the captive portal.
-
@KOM:
If you dont know your certs have been nicked, then they are useful to hackers for a whole year.
Believe me, nobody wants my certs ;D
Fortunately the Let's Encrypt project isn't about only you.
-
But why not just get a https for a year from free like start or there are other places they cost a whole $10 a year.
Let's Encrypt cert is in essence good for as long as the automation runs.
If your so cheap to not want to pay for your ssl.. Just don't bother doing your captive portal via https - or use self signed and deploy that to your users of your captive portal on the your going to have anyway on the http url for the captive portal.
It's not always only about the money. Let's Encrypt cert automation can be very appealing too.
-
Nobody realy knows if the CA you use is trustfull or has a connection to the government or an other party.
Best practice is to generate the private key you wish to use for a certificate locally on an uncompromised machine with a cryptographically secure random number generator. In this scenario the security of the private key is entirely in your hands because the Certificate Signing Request sent to the CA does not contain the private key.
Thats the rub, even if you had an uncompromised machine so many zero days make it possible to obtain such things so if you ever get something like this in your bios http://blog.trendmicro.com/trendlabs-security-intelligence/hacking-team-uses-uefi-bios-rootkit-to-keep-rcs-9-agent-in-target-systems/ well you might just as well throw your system in the bin.
Edit. I'll chuck this link in as its useful which might be of interest. http://www.uefi.org/sites/default/files/resources/UEFI_Secure_Boot_in_Modern_Computer_Security_Solutions_2013.pdf
I see no point to support of this on pfsense to be honest.. Other than maybe captive portal. For the webgui to admin pfsense.. Just use CA on pfsense to create a cert and trust it on your machines you will admin pfsense from. This really should be a really small list of machines!!
I'm no longer using any encryption to access devices or services lan side as I cant inspect the data, any encryption found can be blocked by snort/suricata and investigated. Things like web access to secure websites now take place on a separate device from a linux live cd with no hard drive or other forms of storage and an easily flashable bios.
As to accessing pfsense, switches etc, as above but usernames/password are locally stored so no radius servers for convenience, as its all about minimising the risk of unknowns or to use a more popular phrase, zero days.
If your so cheap to not want to pay for your ssl.. Just don't bother doing your captive portal via https - or use self signed and deploy that to your users of your captive portal on the your going to have anyway on the http url for the captive portal.
Its not so much about being cheap, but trusting the other entities in the supply chain, as it is we have to trust so many people already, reducing that risk just makes sense, which is why we dont trust some people to do some things for us. Others we have to trust when we are incapable of doing something, but where possible its better to do what you can where possible.
-
Let's Encrypt cert is eventually good for as long as the automation runs.
And how often does it look to update? Every day? What if goes to update day before the cert runs out and fails to update for whatever reason - issue on their end, firewall problem on your end, etc.
I don't see the automated updating of ssl to be a good thing to be honest. While I can see this useful on say personal site on some webhost for more users to start using https for their sites. I just do not really see a need off pfsense.. Its a firewall not a WEB SERVER.. Using some automated process to use https for your webgui just seems silly. There should be what a handful of people accessing a firewall gui in the first place.. So why not just issue your own self signed and have those machines used to access it trust the CA that is completely under your control.
Now if you want to use it on your web server behind pfsense - have at it.. I will prob use this on some of my play systems.. Just don't see need/use on my firewall at all.. Especially when that firewall system has a CA..
-
John,
Please go read up on the subject before acting like an expert.
-
I think too a package would be very useful especially when you use squid reverse proxy with Apache and Exchange.
:)
-
I wanted to open a new thread on this but found this one just before posting. ::)
–-
Let's encrypt is a new CA that will begin signing free trusted certificates to the public on 3.12.2015.
The project is founded by the likes of of Mozilla, Akamai, Cisco and the EFF who work together in the Internet Security Research Group (ISRG). [1]
The "catch" is that the certificates have a lifetime of 90 days. Their reasoning behind this is that they want to limit damage from key compromises and they want to encourage automation, which I think makes sense. [2]
These free certificates would be perfect for some pfsense applications like the captive portal or the pfsense web interface.
From what I can tell it has already been implemented in python or javascript so it should run on FreeBSD.[1] https://en.wikipedia.org/wiki/Internet_Security_Research_Group
[2] https://letsencrypt.org/2015/11/09/why-90-days.html -
I wanted to open a new thread on this but found this one just before posting. ::)
–-
Let's encrypt is a new CA that will begin signing free trusted certificates to the public on 3.12.2015.
The project is founded by the likes of of Mozilla, Akamai, Cisco and the EFF who work together in the Internet Security Research Group (ISRG). [1]
The "catch" is that the certificates have a lifetime of 90 days. Their reasoning behind this is that they want to limit damage from key compromises and they want to encourage automation, which I think makes sense. [2]
These free certificates would be perfect for some pfsense applications like the captive portal or the pfsense web interface.
From what I can tell it has already been implemented in python or javascript so it should run on FreeBSD.[1] https://en.wikipedia.org/wiki/Internet_Security_Research_Group
[2] https://letsencrypt.org/2015/11/09/why-90-days.htmlThat sounds good. I think i will try to chnage my pfsense to the letsencrypt ca in the christmas hollidys.
i will post my experiences :) -
NOYB.. And in what point did I say I was an expert? And at what point did I sound like one.. Please explain to me why a FIREWALL with limited access to its webgui by only ADMINS that has its own built in CA already would need/want and automated system to install a cert that expires every 90 days if it can not phone home..
I just do not get it?? I have a cert on my web gui, took all of 2 seconds to create and trust from my different machines I admin pfsense from..
I mentioned already that it might be a good idea for something like captive portal as well.
-
The project is now in public beta.
It seems to be supported on freebsd :
https://letsencrypt.readthedocs.org/en/latest/contributing.html#freebsd
https://letsencrypt.readthedocs.org/en/latest/using.html#installation-and-usage
-
gonzopancho mentioned on reddit that there will be a let's encrypt package in 2.4 ;D
-
Package is there in the nightly, but i don't know how to use it … lol
-
Package is there in the nightly, but i don't know how to use it … lol
It depends on what you want to do with it. For the GUI:
Visit keys tab, make a new entry, click the button to generate a new account key, then click the button to register the key, then save.
Visit the certs tab, make a new cert, enter a hostname and setup a challenge/response method in the SAN list (pick a method, click +, enter the details), then save, then click issue/renew.
What you can use for the challenge depends on what you have available. If your DNS provider for your domain is listed you can probably use one of the DNS update methods, or if your server supports RFC2136 you could setup keys for the TXT records it wants to make and use the nsupdate option (this is what I prefer to do).
I would advise against attempting the webroot method directly on the firewall. You could port forward port 80 on the firewall's WAN address to a local web server and then use the webroot FTP option perhaps if you don't have any other choice.
Once you have managed to get a cert issued, go to System > Advanced, pick it for the GUI cert. Edit the cert entry in the ACME package and setup a new action for a shell command to run /etc/rc.restart_webgui, save again. Visit the general settings tab and check the box, then save.
If you want to use it for something other than the GUI, repeat the process but pick it wherever you need to use it instead (e.g. haproxy), though your update method may vary for that.
I'll write up a more thorough doc on it eventually.
-
@KOM:
90 days?? Geez, why waste time with that when you can get a freebie from StartSSL that's good for a year.
The let's encrypt script can be put on cron, and automate every 90 days with a new cert.
It's done regularly on linux webservers.
-
@KOM:
90 days?? Geez, why waste time with that when you can get a freebie from StartSSL that's good for a year.
One reason may be that the newest version of Google Chrome is no longer trusting StartSSL, due to their parent company doing some shady things.
-
Package is there in the nightly, but i don't know how to use it … lol
It depends on what you want to do with it. For the GUI:
Visit keys tab, make a new entry, click the button to generate a new account key, then click the button to register the key, then save.
Visit the certs tab, make a new cert, enter a hostname and setup a challenge/response method in the SAN list (pick a method, click +, enter the details), then save, then click issue/renew.
What you can use for the challenge depends on what you have available. If your DNS provider for your domain is listed you can probably use one of the DNS update methods, or if your server supports RFC2136 you could setup keys for the TXT records it wants to make and use the nsupdate option (this is what I prefer to do).
I would advise against attempting the webroot method directly on the firewall. You could port forward port 80 on the firewall's WAN address to a local web server and then use the webroot FTP option perhaps if you don't have any other choice.
Once you have managed to get a cert issued, go to System > Advanced, pick it for the GUI cert. Edit the cert entry in the ACME package and setup a new action for a shell command to run /etc/rc.restart_webgui, save again. Visit the general settings tab and check the box, then save.
If you want to use it for something other than the GUI, repeat the process but pick it wherever you need to use it instead (e.g. haproxy), though your update method may vary for that.
I'll write up a more thorough doc on it eventually.
awesome.. i had my dns with namecheap & wasnt able to figure out how to do NSUpdate with them. so moved it to cloudflare & it worked.
If you use cloudflare make sure the dns uses cloudflare DNS only & has a grey cloud. Grey Cloud: Records that display a grey cloud icon will bypass Cloudflare, using only Cloudflare DNS
if you have an orange cloud the auth fails. you can re-enable it after the cert is issued
-
If someone really insists on using a local webroot.
1/ Install HAproxy package.
2/ Put this to HAProxy > Files (Type - Lua script, Name: acme-http01-webroot.lua)(or download from here)
-- ACME http-01 domain validation plugin for Haproxy 1.6+ -- copyright (C) 2015 Jan Broer -- acme = {} acme.version = "0.1.1" -- -- Configuration -- -- When HAProxy is *not* configured with the 'chroot' option you must set an absolute path here and pass -- that as 'webroot-path' to the letsencrypt client acme.conf = { ["non_chroot_webroot"] = "" } -- -- Startup -- acme.startup = function() core.Info("[acme] http-01 plugin v" .. acme.version); end -- -- ACME http-01 validation endpoint -- acme.http01 = function(applet) local response = "" local reqPath = applet.path local src = applet.sf:src() local token = reqPath:match( ".+/(.*)$" ) if token then token = sanitizeToken(token) end if (token == nil or token == '') then response = "bad request\n" applet:set_status(400) core.Warning("[acme] malformed request (client-ip: " .. tostring(src) .. ")") else auth = getKeyAuth(token) if (auth:len() >= 1) then response = auth .. "\n" applet:set_status(200) core.Info("[acme] served http-01 token: " .. token .. " (client-ip: " .. tostring(src) .. ")") else response = "resource not found\n" applet:set_status(404) core.Warning("[acme] http-01 token not found: " .. token .. " (client-ip: " .. tostring(src) .. ")") end end applet:add_header("Server", "haproxy/acme-http01-authenticator") applet:add_header("Content-Length", string.len(response)) applet:add_header("Content-Type", "text/plain") applet:start_response() applet:send(response) end -- -- strip chars that are not in the URL-safe Base64 alphabet -- see https://github.com/letsencrypt/acme-spec/blob/master/draft-barnes-acme.md -- function sanitizeToken(token) _strip="[^%a%d%+%-%_=]" token = token:gsub(_strip,'') return token end -- -- get key auth from token file -- function getKeyAuth(token) local keyAuth = "" local path = acme.conf.non_chroot_webroot .. "/.well-known/acme-challenge/" .. token local f = io.open(path, "rb") if f ~= nil then keyAuth = f:read("*all") f:close() end return keyAuth end core.register_init(acme.startup) core.register_service("acme-http01", "http", acme.http01)
3/ Create a very simple http frontend on WAN address, port 80.
4/ Use this for your certificate(s) in ACME package:
[EDIT: The image host originally used in this post is dead. Fixed using cached copies of the images on another host -jimp]
-
Is there any chance that this will work with Google Domain's DNS? It doesn't look Google provides a way to create txt records using their Dynamic DNS API.
-
What you can use for the challenge depends on what you have available. If your DNS provider for your domain is listed you can probably use one of the DNS update methods, or if your server supports RFC2136 you could setup keys for the TXT records it wants to make and use the nsupdate option (this is what I prefer to do)..
I've been trying to get the manual method to work with he.net, but can't figure out how to generate the TXT key - While I know I need to add it manually to my DNS, is the generation of the key included in your package or is there a manual step required here?
/SJ
-
What you can use for the challenge depends on what you have available. If your DNS provider for your domain is listed you can probably use one of the DNS update methods, or if your server supports RFC2136 you could setup keys for the TXT records it wants to make and use the nsupdate option (this is what I prefer to do)..
I've been trying to get the manual method to work with he.net, but can't figure out how to generate the TXT key - While I know I need to add it manually to my DNS, is the generation of the key included in your package or is there a manual step required here?
/SJ
That was the first method I tested. Define the domain name entry and then click issue/renew. In the green output it tells you what the content of the record should be. Add it to DNS and then wait 2-3 minutes to be sure the record is available, then click issue/renew again.
-
Hello,
new user here. Fresh pfsense install update to 2.3.2_1
I've installed the acme package but i have some problems with Route 53 Dns validation
The output say that:[Mon Feb 6 17:24:12 CET 2017] Registering account [Mon Feb 6 17:24:13 CET 2017] Already registered [Mon Feb 6 17:24:14 CET 2017] Update success. [Mon Feb 6 17:24:14 CET 2017] Single domain='test.sanitazedomain.it' [Mon Feb 6 17:24:14 CET 2017] Getting domain auth token for each domain [Mon Feb 6 17:24:14 CET 2017] Getting webroot for domain='test.sanitazedomain.it' [Mon Feb 6 17:24:14 CET 2017] _w='dns_aws' [Mon Feb 6 17:24:14 CET 2017] Getting new-authz for domain='test.sanitazedomain.it' [Mon Feb 6 17:24:15 CET 2017] The new-authz request is ok. [Mon Feb 6 17:24:16 CET 2017] Found domain api file: /usr/local/pkg/acme/dnsapi/dns_aws.sh [color][Mon Feb 6 17:24:18 CET 2017] Error add txt for domain:_acme-challenge.test.sanitazedomain.it[/color] [Mon Feb 6 17:24:18 CET 2017] Please check log file for more details: /tmp/acme/test/acme_issuecert.log
i checked the log and seem to fail at curl command to retriete http.header.
If i open HTTP.HEADER file in the acme domain folder (test in this case) i get:HTTP/1.1 505 HTTP Version not supported Date: Mon, 06 Feb 17 16:24:18 GMT Connection: close x-amz-id-2: 1rjTvEvOKQpJ5zruKVbddXvS15q4+I1y/+r2qirC9S8MYXm1esOQYwkOscLruZW8zzvK0+WY8BOQiy8GvYMu0rx0Uwq8WqlH x-amz-request-id: 8B82C340F9CA158D Content-Length: 0
any hint? Aws access ID and secret key seems ok. I've tried also to get full access to this IAM user to Route53
-
Route53 made some change to their service in the last few days that might have broken this client. We've had at least one other report of Route53 dyndns not working in general (not related to acme). Odds are the route53 script needs updated to match their new API/methods.
-
Thank you.
Just my luck. Every time i try something new, something is broken since hours or days ::) -
Hello, cert BFU here, so sorry if I won't make much sense…
Is it somehow possible to continue with certs from previous "issuing"? I have used "acme.sh" script in Ubuntu two months ago, sucefully got some acme-challenge TXT values for my (sub)domains, which I have added manually to my DNS configuration and on the second run of "acme.sh" couple files were generated (.cer, .key, ...).I have sucesfully added generated .cer to HAproxy on my pfSense and it is now serving me my https websites through HAproxy and it was my undestanding that when the time comes I would just have to do "acme.sh --renew -d mydomain.com" to regenerate certs and manually replace cert on HAproxy.
I wanted to automate this using this pfSense package. Is it possible to continue with started process, or do I have to generate new set of TXT values and replace them at my DNS config again?
I have tried to put content of my .key file into "Account keys" tab and define same domainname on Certificates tab with Method: DNS-manual, but attempt to "Renew" ends with green "mydomain.com is not a issued domain, skip" message.
Am I doing it all wrong?
-
The TXT records are only valid for a few days and then they expire – you'd have to remake them when it's time to renew anyhow.
If you use the exact same list of SANs from your original cert, LE will allow it can will consider it a reissue. If you change the the list of SANs, it's treated as a new certificate. (Not too important unless you're close to their rate limits...)
-
That was the first method I tested. Define the domain name entry and then click issue/renew. In the green output it tells you what the content of the record should be. Add it to DNS and then wait 2-3 minutes to be sure the record is available, then click issue/renew again.
Thanks for the response! Oddly, when I configure the manual method I get both an issue and a renew button, rather than the joined button I get if it was set to webroot, but I think that's a minor detail.
My output, however, holds no TXT entry, which is why I was getting confused
xxx.net Renewing certificateaccount: xxx-key server: letsencrypt-production /usr/local/pkg/acme/acme.sh --issue -d 'host.xxx.net' --home '/tmp/acme/xxx.net/' --accountconf '/tmp/acme/xxx.net/accountconf.conf' --force --reloadCmd '/tmp/acme/xxx.net/reloadcmd.sh' --dns '' --log-level 3 --log '/tmp/acme/xxx.net/acme_issuecert.log' Array ( [path] => /etc:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin/ [PATH] => /etc:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin/ ) [Mon Feb 6 21:03:55 UTC 2017] Single domain='host.xxx.net' [Mon Feb 6 21:03:55 UTC 2017] Getting domain auth token for each domain [Mon Feb 6 21:03:55 UTC 2017] Getting webroot for domain='host.xxx.net' [Mon Feb 6 21:03:55 UTC 2017] _w [Mon Feb 6 21:03:55 UTC 2017] Getting new-authz for domain='host.xxx.net' [Mon Feb 6 21:03:59 UTC 2017] The new-authz request is ok. [Mon Feb 6 21:03:59 UTC 2017] Verifying:host.xxx.net [Mon Feb 6 21:04:02 UTC 2017] Pending [Mon Feb 6 21:04:05 UTC 2017] host.xxx.net:Verify error:Could not connect to host.xxx.net [Mon Feb 6 21:04:05 UTC 2017] Please check log file for more details: /tmp/acme/xxx.net/acme_issuecert.log
-
I found a problem with it and pushed a fix a couple hours ago. Update the package to 0.1.8 when you see it. I put in fixes for SFTP Webroot and DNS-Manual today.
https://github.com/pfsense/FreeBSD-ports/commit/a8952d9a67c674e521f75f5c3e61d879f89d43a4
https://github.com/pfsense/FreeBSD-ports/commit/9bda224f04c361836ebd7ebf1992de20d504487bThe new package won't show up for 2.3.3 or 2.4 until new snapshots are built.
-
I found a problem with it and pushed a fix a couple hours ago. Update the package to 0.1.8 when you see it. I put in fixes for SFTP Webroot and DNS-Manual today.
https://github.com/pfsense/FreeBSD-ports/commit/a8952d9a67c674e521f75f5c3e61d879f89d43a4
https://github.com/pfsense/FreeBSD-ports/commit/9bda224f04c361836ebd7ebf1992de20d504487bThe new package won't show up for 2.3.3 or 2.4 until new snapshots are built.
I see the update and giving it a shot now! Shall report back in a jiffy.
-
Couple Caveats so far:
-
When you click Issue and receive the TXT record (that works lovely now thank you!) make sure you hit Renew after you have added the TXT record to your domain, otherwise you will simply generate a new TXT record.
-
Somehow it has generated the cert as self-signed and not via Let's Encrypt which is throwing errors in the browser. Might be because of the update, going to start from scratch and see how it goes.
EDIT: So I still had my Startcom Root CA on the firewall, which seemed to mean the Let's Encrypt CA wouldn't install. After removing that, and reissuing the cert, it's all going swimmingly.
Cheers jimp! Great little package!
-
-
I found a problem with it and pushed a fix a couple hours ago. Update the package to 0.1.8 when you see it. I put in fixes for SFTP Webroot and DNS-Manual today.
https://github.com/pfsense/FreeBSD-ports/commit/a8952d9a67c674e521f75f5c3e61d879f89d43a4
https://github.com/pfsense/FreeBSD-ports/commit/9bda224f04c361836ebd7ebf1992de20d504487bThe new package won't show up for 2.3.3 or 2.4 until new snapshots are built.
got an update to 0.1.9
github says "Sorry, this commit history is taking too long to generate."
any tip on what's new/fixed in this release?
-
0.1.9 was this PR: https://github.com/pfsense/FreeBSD-ports/pull/296/files
-
FYI-
acme pkg version 0.1.10 = Fix formatting of nsupdate options
acme pkg version 0.1.11 = Add key type and algorithm selection to nsupdate, other fixes to nsupdate
acme pkg version 0.1.12 = Add standalone HTTP and TLS server methodsRe: Standalone mode: For security reasons, Let's Encrypt requires 80 for HTTP and 443 for TLS checks. If you bind to any other port you must forward port 80/443 to that other port or the check will fail. And it goes without saying that you must allow that traffic to the firewall with rules.
Perhaps in the future it could be fancy and add a temporary rule to pass traffic to the port, but for now you'll have to add one by hand. And you'll probably want to shut down the rule afterward unless you really want to leave that port open all the time.
-
Two small bugs/improvement requests.
1. If you click on the questionmark in the acme plugin page - it sends you to a non existing page.
2. For each entry in the domain san list - the password is shown in clear text even in the summary - it would be better to mask it -
If anyone is interested I wrote a how-to post on using this Let's Encrypt package.
https://blog.artooro.com/2017/02/16/quick-easy-lets-encrypt-setup-on-pfsense-using-acme/It works great already. Thanks for building it!
The biggest improvement I'd like for myself is the ability to select whether an "Actions list" command is executed before or after the renewal/issue. This way a command could be used to create a firewall rule that allows ACME to verify, and then the rule can be automatically removed when finished.
-
I'm trying to wrap my head around making LetsEncrypt work with my HAProxy setup so that renewals go through automatically, regardless of which domain validation requests come through.
Here's the current setup on pfSense:
Ports 80 and 443 open on firewall
HAProxy reverse proxy routing both HTTP and HTTPS requests to appropriate backend servers. Only simply routingHTTP/SNI SSL passthrough, no SSL termination at this point.I'd like to take advantage of the LetsEncrypt package, and was thinking of using the Standalone HTTP server option with a non-standard port, and then having HAProxy route the validation requests (carrying forward with automated renewals, obviously) to this Standalone Server instance.
Something like ACL (if /.well-known/blah blah blah/) use backend LetsEncrypt-standalone-server.I am assuming that this provides a better (more secure?) option than using wwwroot, since that's the gist I got from this thread.
Now, my first question is about whether I need to set the firewall up to allow any additional ports through (say, to the Standalone Server's port from ??? where), or is this rendered unnecessary because both the proxy and the Standalone Server are on the same host (the pfSense router)?
Secondly, when creating the backend server entry for HAProxy to route to for the Standalone Server (the LetsEncrypt response/validation server), what do I use for this? 127.0.0.1:port or the router's address? 192.168.1.1:port?
Thank you for your time.
-
Possible bug, one { too much, Please look at:
https://forum.pfsense.org/index.php?topic=125946.0
Thanks
-
I have installed the package on many machines but some of them are unable to store account keys it seems.
They can be generated but clicking on "save" does not seem to make them permanently available.Is there anything I can do about it?
Regards,
Julian