added option to pass environment variables over sudo

Newer versions of openssl seem to send the verify outout to stdout instead of
stderr in the past. Ignore that output when retrieving altnames.

.github/FUNDING.yml

@@ -0,0 +1,2 @@
+github: lukas2511
+custom: ["https://paypal.me/lukas2511", "http://www.amazon.de/registry/wishlist/1TUCFJK35IO4Q"]

CHANGELOG

@@ -1,6 +1,70 @@
 # Change Log
 This file contains a log of major changes in dehydrated
 
+## [x.x.x] - xxxx-xx-xx
+## Added
+- New config variable `DEHYDRATED_SUDO_ENV` to allow passing environment variables over sudo calls
+
+## [0.7.1] - 2022-10-31
+## Changed
+- `--force` no longer forces domain name revalidation by default, a new argument `--force-validation` has been added for that
+- Added support for EC secp521r1 algorithm (works with e.g. zerossl)
+- `EC PARAMETERS` are no longer written to privkey.pem (didn't seem necessary and was causing issues with various software)
+
+## Fixed
+- Requests resulting in `badNonce` errors are now automatically retried (fixes operation with LE staging servers)
+- Deprecated `egrep` usage has been removed
+
+## Added
+- Implemented EC for account keys
+- Domain list now also read from domains.txt.d subdirectory (behaviour might change, see docs)
+- Implemented RFC 8738 (validating/signing certificates for IP addresses instead of domain names) support (this will not work with most public CAs, if any!)
+
+## [0.7.0] - 2020-12-10
+## Added
+- Support for external account bindings
+- Special support for ZeroSSL
+- Support presets for some CAs instead of requiring URLs
+- Allow requesting preferred chain (`--preferred-chain`)
+- Added method to show CAs current terms of service (`--display-terms`)
+- Allow setting path to domains.txt using cli arguments (`--domains-txt`)
+- Added new cli command `--cleanupdelete` which deletes old files instead of archiving them
+
+## Fixed
+- No more silent failures on broken hook-scripts
+- Better error-handling with KEEP_GOING enabled
+- Check actual order status instead of assuming it's valid
+- Don't include keyAuthorization in challenge validation (RFC compliance)
+
+## Changed
+- Using EC secp384r1 as default certificate type
+- Use JSON.sh to parse JSON
+- Use account URL instead of account ID (RFC compliance)
+- Dehydrated now has a new home: https://github.com/dehydrated-io/dehydrated
+- Added `OCSP_FETCH` and `OCSP_DAYS` to per-certificate configurable options
+- Cleanup now also removes dangling symlinks
+
+## [0.6.5] - 2019-06-26
+## Fixed
+- Fixed broken APIv1 compatibility from last update
+
+## [0.6.4] - 2019-06-25
+## Changed
+- Fetch account ID from Location header instead of account json
+
+## [0.6.3] - 2019-06-25
+## Changed
+- OCSP refresh interval is now configurable
+- Implemented POST-as-GET
+- Call exit_hook on errors (with error-message as first parameter)
+
+## Added
+- Initial support for tls-alpn-01 validation
+- New hook: sync_cert (for syncing certificate files to disk, see example hook description)
+
+## Fixes
+- Fetch account information after registration to avoid missing account id
+
 ## [0.6.2] - 2018-04-25
 ## Added
 - New deploy_ocsp hook

LICENSE

@@ -1,6 +1,6 @@
 The MIT License (MIT)
 
-Copyright (c) 2015-2018 Lukas Schauer
+Copyright (c) 2015-2021 Lukas Schauer
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

README.md

@@ -1,6 +1,6 @@
 # dehydrated [![Donate](https://img.shields.io/badge/Donate-PayPal-green.svg)](https://www.paypal.com/cgi-bin/webscr?cmd=_s-xclick&hosted_button_id=23P9DSJBTY7C8)
 
-![](docs/logo.jpg)
+![](docs/logo.png)
 
 Dehydrated is a client for signing certificates with an ACME-server (e.g. Let's Encrypt) implemented as a relatively simple (zsh-compatible) bash-script.
 This client supports both ACME v1 and the new ACME v2 including support for wildcard certificates!
@@ -14,6 +14,7 @@ Current features:
 - Signing of a custom CSR (either standalone or completely automated using hooks!)
 - Renewal if a certificate is about to expire or defined set of domains changed
 - Certificate revocation
+- and lots more..
 
 Please keep in mind that this software, the ACME-protocol and all supported CA servers out there are relatively young and there might be a few issues. Feel free to report any issues you find with this script or contribute by submitting a pull request,
 but please check for duplicates first (feel free to comment on those to get things rolling).
@@ -49,12 +50,15 @@ Default command: help
 
 Commands:
  --version (-v)                   Print version information
+ --display-terms                  Display current terms of service
  --register                       Register account key
  --account                        Update account contact information
  --cron (-c)                      Sign/renew non-existent/changed/expiring certificates.
  --signcsr (-s) path/to/csr.pem   Sign a given CSR, output CRT on stdout (advanced usage)
  --revoke (-r) path/to/cert.pem   Revoke specified certificate
+ --deactivate                     Deactivate account
  --cleanup (-gc)                  Move unused certificate files to archive directory
+ --cleanup-delete (-gcd)          Deletes (!) unused certificate files
  --help (-h)                      Show help text
  --env (-e)                       Output configuration variables for use in other scripts
 
@@ -64,38 +68,27 @@ Parameters:
  --ipv4 (-4)                      Resolve names to IPv4 addresses only
  --ipv6 (-6)                      Resolve names to IPv6 addresses only
  --domain (-d) domain.tld         Use specified domain name(s) instead of domains.txt entry (one certificate!)
+ --ca url/preset                  Use specified CA URL or preset
  --alias certalias                Use specified name for certificate directory (and per-certificate config) instead of the primary domain (only used if --domain is specified)
  --keep-going (-g)                Keep going after encountering an error while creating/renewing multiple certificates in cron mode
- --force (-x)                     Force renew of certificate even if it is longer valid than value in RENEW_DAYS
+ --force (-x)                     Force certificate renewal even if it is not due to expire within RENEW_DAYS
+ --force-validation               Force revalidation of domain names (used in combination with --force)
  --no-lock (-n)                   Don't use lockfile (potentially dangerous!)
  --lock-suffix example.com        Suffix lockfile name with a string (useful for with -d)
  --ocsp                           Sets option in CSR indicating OCSP stapling to be mandatory
  --privkey (-p) path/to/key.pem   Use specified private key instead of account key (useful for revocation)
+ --domains-txt path/to/domains.txt Use specified domains.txt instead of default/configured one
  --config (-f) path/to/config     Use specified config file
  --hook (-k) path/to/hook.sh      Use specified script for hooks
+ --preferred-chain issuer-cn      Use alternative certificate chain identified by issuer CN
  --out (-o) certs/directory       Output certificates into the specified directory
- --challenge (-t) http-01|dns-01  Which challenge should be used? Currently http-01 and dns-01 are supported
+ --alpn alpn-certs/directory      Output alpn verification certificates into the specified directory
+ --challenge (-t) http-01|dns-01|tls-alpn-01 Which challenge should be used? Currently http-01, dns-01, and tls-alpn-01 are supported
  --algo (-a) rsa|prime256v1|secp384r1 Which public key algorithm should be used? Supported: rsa, prime256v1 and secp384r1
 ```
 
-## Donate
+## Chat
 
-I'm a student hacker with a few (unfortunately) quite expensive hobbies (self-hosting, virtualization clusters, routing,
-high-speed networking, embedded hardware, etc.).
-I'm really having fun playing around with hard- and software and I'm steadily learning new things.
-Without those hobbies I probably would never have started working on dehydrated to begin with :)
+Dehydrated has an official IRC-channel `#dehydrated` on libera.chat that can be used for general discussion and suggestions.
 
-I'd really appreciate if you could [donate a bit of money](https://www.paypal.com/cgi-bin/webscr?cmd=_s-xclick&hosted_button_id=23P9DSJBTY7C8)
-so I can buy cool stuff (while still being able to afford food :D).  
-
-If you have hardware laying around that you think I'd enjoy playing with (e.g. decommissioned but still modern-ish servers,
-10G networking hardware, enterprise grade routers or APs, interesting ARM/MIPS boards, etc.) and that you would be willing
-to ship to me please contact me at `donations@dehydrated.io` or on Twitter [@lukas2511](https://twitter.com/lukas2511).
-
-If you want your name to be added to the [donations list](https://dehydrated.io/donations.html) please add a note or send me an
-email `donations@dehydrated.io`. I respect your privacy and won't publish your name without permission.
-
-Other ways of donating:
- - [My Amazon Wishlist](http://www.amazon.de/registry/wishlist/1TUCFJK35IO4Q)
- - Monero: 4Kkf4tF4r9DakxLj37HDXLJgmpVfQoFhT7JLDvXwtUZZMTbsK9spsAPXivWPAFcDUj6jHhY8hJSHX8Cb8ndMhKeQHPSkBZZiK89Fx8NTHk
- - Bitcoin: 12487bHxcrREffTGwUDnoxF1uYxCA7ztKK
+The channel can also be accessed with Matrix using the official libera.chat bridge at `#dehydrated:libera.chat`.

docs/acme-v1.md

@@ -0,0 +1,19 @@
+## (Future) Removal of API version 1
+
+The ACME API version 1 was never really standardized and was only supported by Let's Encrypt. Even though the protocol specification was public,
+it wasn't really friendly to be integrated into existing CA systems so initial adoption was basically non-existant.
+
+ACME version 2 is being designed to overcome these issues by becoming an official IETF standard and supporting a more traditional approach of account
+and order management in the backend, making it friendlier to integrate into existing systems centered around those. It has since become a semi-stable IETF
+standard draft which only ever got two breaking changes, Content-Type enforcement and `POST-as-GET`, the latter being announced in October 2018 to be enforced
+by November 2019. See https://datatracker.ietf.org/wg/acme/documents/ for a better insight into the draft and its changes.
+
+Next to backend changes that many users won't really care about ACME v2 has all of the features ACME v1 had, but also some additional new features like
+e.g. support for [wildcard certificates](domains_txt.md#wildcards).
+
+Since ACME v2 is basically to be considered stable and ACME v1 has no real benefits over v2, there doesn't seem to be much of a reason to keep the old
+protocol around, but since there actually are a few Certificate Authorities and resellers that implemented the v1 protocol and didn't yet make the change
+to v2, so dehydrated still supports the old protocol for now.
+
+Please keep in mind that support for the old ACME protocol version 1 might get removed at any point of bigger inconvenience, e.g. on code changes that
+would require a lot of work or ugly workarounds to keep both versions supported.

docs/dns-verification.md

@@ -28,4 +28,4 @@ Or when you do have a DNS API, pass the details accordingly to achieve the same
 
 You can delete the TXT record when called with operation `clean_challenge`, when $2 is also the domain name.
 
-Here are some examples: [Examples for DNS-01 hooks](https://github.com/lukas2511/dehydrated/wiki/Examples-for-DNS-01-hooks)
+Here are some examples: [Examples for DNS-01 hooks](https://github.com/dehydrated-io/dehydrated/wiki)

docs/domains_txt.md

@@ -34,6 +34,30 @@ under your `CERTDIR`.
 example.net www.example.net wiki.example.net > certalias
 ```
 
+This allows to set per certificates options. The options you can change are
+explained in [Per Certificate Config](per-certificate-config.md).
+
+If you want to create different certificate types for the same domain
+you can use:
+
+```text
+*.service.example.org service.example.org  > star_service_example_org_rsa
+*.service.example.org service.example.org  > star_service_example_org_ecdsa
+```
+
+Then add a config file `certs/star_service_example_org_rsa/config` with
+the value
+
+```
+KEY_ALGO="rsa"
+```
+
+or respectively
+
+```
+KEY_ALGO="ecdsa"
+```
+
 ### Wildcards
 
 Support for wildcards was added by the ACME v2 protocol.
@@ -70,3 +94,14 @@ This creates two certificates one for `service.example.com` with an
 **Note:** The first certificate is valid for both `service.example.com` and for
 `*.service.example.com` which can be a useful way to create wildcard
 certificates.
+
+### Drop-in directory
+
+If a directory named `domains.txt.d` exists in the same location as
+`domains.txt`, the contents of `*.txt` files in that directory are appended to
+the list of domains, in alphabetical order of the filenames. This is useful for
+automation, as it doesn't require editing an existing file to add new domains.
+
+Warning: Behaviour of this might change as the naming between `domains.txt.d`
+and the `DOMAINS_D` config variable (which is used for per-certificate
+configuration) is a bit confusing.

docs/examples/config

@@ -10,19 +10,24 @@
 # Default values of this config are in comments        #
 ########################################################
 
-# Which user should dehydrated run as? This will be implictly enforced when running as root
+# Which user should dehydrated run as? This will be implicitly enforced when running as root
 #DEHYDRATED_USER=
 
-# Which group should dehydrated run as? This will be implictly enforced when running as root
+# Which group should dehydrated run as? This will be implicitly enforced when running as root
 #DEHYDRATED_GROUP=
 
+# Should dehydrated pass environment variables over sudo?
+#DEHYDRATED_SUDO_ENV="no"
+
 # Resolve names to addresses of IP version only. (curl)
 # supported values: 4, 6
 # default: <unset>
 #IP_VERSION=
 
-# Path to certificate authority (default: https://acme-v02.api.letsencrypt.org/directory)
-#CA="https://acme-v02.api.letsencrypt.org/directory"
+# URL to certificate authority or internal preset
+# Presets: letsencrypt, letsencrypt-test, zerossl, buypass, buypass-test
+# default: letsencrypt
+#CA="letsencrypt"
 
 # Path to old certificate authority
 # Set this value to your old CA value when upgrading from ACMEv1 to ACMEv2 under a different endpoint.
@@ -31,7 +36,7 @@
 # default: https://acme-v01.api.letsencrypt.org/directory
 #OLDCA="https://acme-v01.api.letsencrypt.org/directory"
 
-# Which challenge should be used? Currently http-01 and dns-01 are supported
+# Which challenge should be used? Currently http-01, dns-01 and tls-alpn-01 are supported
 #CHALLENGETYPE="http-01"
 
 # Path to a directory containing additional config files, allowing to override
@@ -40,6 +45,11 @@
 # default: <unset>
 #CONFIG_D=
 
+# Directory for per-domain configuration files.
+# If not set, per-domain configurations are sourced from each certificates output directory.
+# default: <unset>
+#DOMAINS_D=
+
 # Base directory for account key, generated certificates and list of domains (default: $SCRIPTDIR -- uses config directory if undefined)
 #BASEDIR=$SCRIPTDIR
 
@@ -49,6 +59,9 @@
 # Output directory for generated certificates
 #CERTDIR="${BASEDIR}/certs"
 
+# Output directory for alpn verification certificates
+#ALPNCERTDIR="${BASEDIR}/alpn-certs"
+
 # Directory for account keys and registration information
 #ACCOUNTDIR="${BASEDIR}/accounts"
 
@@ -92,7 +105,7 @@
 #PRIVATE_KEY_ROLLOVER="no"
 
 # Which public key algorithm should be used? Supported: rsa, prime256v1 and secp384r1
-#KEY_ALGO=rsa
+#KEY_ALGO=secp384r1
 
 # E-mail to use during the registration (default: <unset>)
 #CONTACT_EMAIL=
@@ -106,6 +119,9 @@
 # Fetch OCSP responses (default: no)
 #OCSP_FETCH="no"
 
+# OCSP refresh interval (default: 5 days)
+#OCSP_DAYS=5
+
 # Issuer chain cache directory (default: $BASEDIR/chains)
 #CHAINCACHE="${BASEDIR}/chains"
 
@@ -114,3 +130,6 @@
 
 # ACME API version (default: auto)
 #API=auto
+
+# Preferred issuer chain (default: <unset> -> uses default chain)
+#PREFERRED_CHAIN=

docs/examples/domains.txt

@@ -24,6 +24,15 @@ example.net www.example.net > certalias
 # NOTE: It is a certificate for 'service.example.org'
 *.service.example.org service.example.org  > star_service_example_org
 
+# Optionally you can also append the certificate algorithm here to create
+# multiple certificate types for the same domain.
+#
+# This allows to set per certificates options. How to do this is
+# explained in [domains.txt documentation](domains_txt.md).
+#
+*.service.example.org service.example.org  > star_service_example_org_rsa
+*.service.example.org service.example.org  > star_service_example_org_ecdsa
+
 # Create a certificate for 'service.example.net' with an alternative name of
 # '*.service.example.net' (which is a wildcard domain) and store it in the
 # directory ${CERTDIR}/service.example.net

docs/examples/hook.sh

@@ -37,6 +37,32 @@ clean_challenge() {
   # printf 'server 127.0.0.1\nupdate delete _acme-challenge.%s TXT "%s"\nsend\n' "${DOMAIN}" "${TOKEN_VALUE}" | nsupdate -k /var/run/named/session.key
 }
 
+sync_cert() {
+  local KEYFILE="${1}" CERTFILE="${2}" FULLCHAINFILE="${3}" CHAINFILE="${4}" REQUESTFILE="${5}"
+
+  # This hook is called after the certificates have been created but before
+  # they are symlinked. This allows you to sync the files to disk to prevent
+  # creating a symlink to empty files on unexpected system crashes.
+  #
+  # This hook is not intended to be used for further processing of certificate
+  # files, see deploy_cert for that.
+  #
+  # Parameters:
+  # - KEYFILE
+  #   The path of the file containing the private key.
+  # - CERTFILE
+  #   The path of the file containing the signed certificate.
+  # - FULLCHAINFILE
+  #   The path of the file containing the full certificate chain.
+  # - CHAINFILE
+  #   The path of the file containing the intermediate certificate(s).
+  # - REQUESTFILE
+  #   The path of the file containing the certificate signing request.
+
+  # Simple example: sync the files before symlinking them
+  # sync "${KEYFILE}" "${CERTFILE}" "${FULLCHAINFILE}" "${CHAINFILE}" "${REQUESTFILE}"
+}
+
 deploy_cert() {
   local DOMAIN="${1}" KEYFILE="${2}" CERTFILE="${3}" FULLCHAINFILE="${4}" CHAINFILE="${5}" TIMESTAMP="${6}"
 
@@ -151,7 +177,7 @@ generate_csr() {
   # This hook is called before any certificate signing operation takes place.
   # It can be used to generate or fetch a certificate signing request with external
   # tools.
-    # The output should be just the cerificate signing request formatted as PEM.
+  # The output should be just the certificate signing request formatted as PEM.
   #
   # Parameters:
   # - DOMAIN
@@ -178,13 +204,17 @@ startup_hook() {
 }
 
 exit_hook() {
+  local ERROR="${1:-}"
+
   # This hook is called at the end of the cron command and can be used to
   # do some final (cleanup or other) tasks.
-
-  :
+  #
+  # Parameters:
+  # - ERROR
+  #   Contains error message if dehydrated exits with error
 }
 
 HANDLER="$1"; shift
-if [[ "${HANDLER}" =~ ^(deploy_challenge|clean_challenge|deploy_cert|deploy_ocsp|unchanged_cert|invalid_challenge|request_failure|generate_csr|startup_hook|exit_hook)$ ]]; then
+if [[ "${HANDLER}" =~ ^(deploy_challenge|clean_challenge|sync_cert|deploy_cert|deploy_ocsp|unchanged_cert|invalid_challenge|request_failure|generate_csr|startup_hook|exit_hook)$ ]]; then
   "$HANDLER" "$@"
 fi

docs/import-from-official-client.md

@@ -1,3 +0,0 @@
-# Import
-
-If you want to import existing keys from the official letsencrypt client have a look at [Import from official letsencrypt client](https://github.com/lukas2511/dehydrated/wiki/Import-from-official-letsencrypt-client).

docs/man/dehydrated.1

@@ -20,8 +20,8 @@ Dehydrated will notify if no account is configured. Run with \fB--register
 
 Next, all domain names must be provided in domains.txt. The format is line
 based: If the file contains two lines "example.com" and "example.net",
-Dehydrated will request two certificate, one for "example.com" and the other
-for "example.net". A single line while "example.com example.net" will request a
+dehydrated will request two certificate, one for "example.com" and the other
+for "example.net". A single line containing "example.com example.net" will request a
 single certificate valid for both "example.net" and "example.com" through the \fISubject
 Alternative Name\fR (SAN) field.
 
@@ -106,7 +106,7 @@ Keep going after encountering an error while creating/renewing multiple
 certificates in cron mode
 .TP
 .BR \-\-force ", " \-x
-Force renew of certificate even if it is longer valid than value in RENEW_DAYS
+Force certificate renewal even if it is not due to expire within RENEW_DAYS
 .TP
 .BR \-\-no\-lock ", " \-n
 Don't use lockfile (potentially dangerous!)

docs/per-certificate-config.md

@@ -7,15 +7,19 @@ To use this feature create a `config` file in the certificates output directory
 Currently supported options:
 
 - PRIVATE_KEY_RENEW
+- PRIVATE_KEY_ROLLOVER
 - KEY_ALGO
 - KEYSIZE
 - OCSP_MUST_STAPLE
+- OCSP_FETCH
+- OCSP_DAYS
 - CHALLENGETYPE
 - HOOK
 - HOOK_CHAIN
 - WELLKNOWN
 - OPENSSL_CNF
 - RENEW_DAYS
+- PREFERRED_CHAIN
 
 ## DOMAINS_D
 

docs/staging.md

@@ -8,10 +8,7 @@ you will quickly hit these limits and find yourself locked out.
 To avoid this, please set the CA property to the Let’s Encrypt staging server URL in your config file:
 
 ```bash
-CA="https://acme-staging.api.letsencrypt.org/directory"
+CA="https://acme-staging-v02.api.letsencrypt.org/directory"
 ```
 
-# ACMEv2 staging
-
-You can use `CA="https://acme-staging-v02.api.letsencrypt.org/directory"` to test dehydrated with
-the ACMEv2 staging endpoint.
+Alternatively you can define the CA using the CLI argument `--ca letsencrypt-test` (`letsencrypt-test` is an integrated preset-CA corresponding to the URL above).

docs/tls-alpn.md

@@ -0,0 +1,124 @@
+# TLS-ALPN-01
+
+With `tls-alpn-01`-type verification Let's Encrypt (or the ACME-protocol in general) is checking if you are in control of a domain by accessing
+your webserver using a custom ALPN and expecting a specially crafted TLS certificate containing a verification token.
+It will do that for any (sub-)domain you want to sign a certificate for.
+
+Dehydrated generates the required verification certificates, but the delivery is out of its scope.
+
+### Example lighttpd config
+
+lighttpd can be configured to recognize ALPN `acme-tls/1` and to respond to such
+requests using the specially crafted TLS certificates generated by dehydrated.
+Configure lighttpd and dehydrated to use the same path for these certificates.
+(Be sure to allow read access to the user account under which the lighttpd
+server is running.) `mkdir -p /etc/dehydrated/alpn-certs`
+
+lighttpd.conf:
+```
+ssl.acme-tls-1 = "/etc/dehydrated/alpn-certs"
+```
+
+When renewing certificates, specify `-t tls-alpn-01` and `--alpn /etc/dehydrated/alpn-certs` to dehydrated, e.g.
+```
+dehydrated -t tls-alpn-01 --alpn /etc/dehydrated/alpn-certs -c --out /etc/lighttpd/certs -d www.example.com
+# gracefully reload lighttpd to use the new certificates by sending lighttpd pid SIGUSR1
+systemctl reload lighttpd
+```
+
+### Example nginx config
+
+On an nginx tcp load-balancer you can use the `ssl_preread` module to map a different port for acme-tls
+requests than for e.g. HTTP/2 or HTTP/1.1 requests.
+
+Your config should look something like this:
+
+```nginx
+stream {
+  map $ssl_preread_alpn_protocols $tls_port {
+    ~\bacme-tls/1\b 10443;
+    default 443;
+  }
+
+  server {
+    listen 443;
+    listen [::]:443;
+    proxy_pass 10.13.37.42:$tls_port;
+    ssl_preread on;
+  }
+}
+```
+
+That way https requests are forwarded to port 443 on the backend server, and acme-tls/1 requests are
+forwarded to port 10443.
+
+In the future nginx might support internal routing based on custom ALPNs, but for now you'll have to
+use a custom responder for the alpn verification certificates (see below).
+
+### Example responder
+
+I hacked together a simple responder in Python, it might not be the best, but it works for me:
+
+```python
+#!/usr/bin/env python3
+
+import ssl
+import socketserver
+import threading
+import re
+import os
+
+ALPNDIR="/etc/dehydrated/alpn-certs"
+PROXY_PROTOCOL=False
+
+FALLBACK_CERTIFICATE="/etc/ssl/certs/ssl-cert-snakeoil.pem"
+FALLBACK_KEY="/etc/ssl/private/ssl-cert-snakeoil.key"
+
+class ThreadedTCPServer(socketserver.ThreadingMixIn, socketserver.TCPServer):
+    pass
+
+class ThreadedTCPRequestHandler(socketserver.BaseRequestHandler):
+    def create_context(self, certfile, keyfile, first=False):
+        ssl_context = ssl.create_default_context(ssl.Purpose.CLIENT_AUTH)
+        ssl_context.set_ciphers('ECDHE+AESGCM')
+        ssl_context.set_alpn_protocols(["acme-tls/1"])
+        ssl_context.options |= ssl.OP_NO_TLSv1 | ssl.OP_NO_TLSv1_1
+        if first:
+            ssl_context.set_servername_callback(self.load_certificate)
+        ssl_context.load_cert_chain(certfile=certfile, keyfile=keyfile)
+        return ssl_context
+
+    def load_certificate(self, sslsocket, sni_name, sslcontext):
+        print("Got request for %s" % sni_name)
+        if not re.match(r'^(([a-zA-Z]{1})|([a-zA-Z]{1}[a-zA-Z]{1})|([a-zA-Z]{1}[0-9]{1})|([0-9]{1}[a-zA-Z]{1})|([a-zA-Z0-9][-_.a-zA-Z0-9]{0,61}[a-zA-Z0-9]))\.([a-zA-Z]{2,13}|[a-zA-Z0-9-]{2,30}.[a-zA-Z]{2,3})$', sni_name):
+            return
+
+        certfile = os.path.join(ALPNDIR, "%s.crt.pem" % sni_name)
+        keyfile = os.path.join(ALPNDIR, "%s.key.pem" % sni_name)
+
+        if not os.path.exists(certfile) or not os.path.exists(keyfile):
+            return
+
+        sslsocket.context = self.create_context(certfile, keyfile)
+
+    def handle(self):
+        if PROXY_PROTOCOL:
+            buf = b""
+            while b"\r\n" not in buf:
+                buf += self.request.recv(1)
+
+        ssl_context = self.create_context(FALLBACK_CERTIFICATE, FALLBACK_KEY, True)
+        newsock = ssl_context.wrap_socket(self.request, server_side=True)
+
+if __name__ == "__main__":
+    HOST, PORT = "0.0.0.0", 10443
+
+    server = ThreadedTCPServer((HOST, PORT), ThreadedTCPRequestHandler, bind_and_activate=False)
+    server.allow_reuse_address = True
+    try:
+        server.server_bind()
+        server.server_activate()
+        server.serve_forever()
+    except:
+        server.shutdown()
+```

docs/troubleshooting.md

@@ -30,3 +30,23 @@ If you are using HTTP validation make sure that the path you have configured wit
 To test this create a file (e.g. `test.txt`) in that directory and try opening it with your browser: `http://example.org/.well-known/acme-challenge/test.txt`. Note that if you have an IPv6 address, the challenge connection will be on IPv6. Be sure that you test HTTP connections on both IPv4 and IPv6. Checking the test file in your browser is often not sufficient because the browser just fails over to IPv4.
 
 If you get any error you'll have to fix your web server configuration.
+
+## DNS invalid challenge since dehydrated 0.6.0 / Why are DNS challenges deployed first and verified later?
+
+Since Let's Encrypt (and in general the ACMEv2 protocol) now supports wildcard domains there is a situation where DNS caching can become a problem.
+If somebody wants to validate a certificate with `example.org` and `*.example.org` there are two tokens that have to be deployed on `_acme-challenge.example.org`.
+
+If dehydrated would deploy and verify each token on its own the CA would cache the first token on `_acme-challenge.example.org` and the next challenge would simply fail.
+Let's Encrypt uses your DNS TTL with a max limit of 5 minutes, but this doesn't seem to be part of the ACME protocol, just some LE specific configuration,
+so with other CAs and certain DNS providers who don't allow low TTLs this could potentially take hours.
+
+Since dehydrated now deploys all challenges first that no longer is a problem. The CA will query and cache both challenges, and both authorizations can be validated.
+Some hook-scripts were written in a way that erases the old TXT record rather than adding a new entry, those should be (and many of them already have been) fixed.
+
+There are certain DNS providers which really only allow one TXT record on a domain. This is really odd and you should probably contact your DNS provider and ask them
+to fix this.
+
+If for whatever reason you can't switch DNS providers and your DNS provider only supports one TXT record and doesn't want to fix that you could try splitting your
+certificate into multiple certificates and add a sleep in the `deploy_cert` hook.
+If you can't do that or really don't want to please leave a comment on https://github.com/lukas2511/dehydrated/issues/554,
+if many people are having this unfixable problem I might try to implement a workaround.