i wrote this in response to erica's thread at
https://opensource.eff.org/eff-open-source/pl/xtuemgdti78xfx1hn9jwbrfrjy
this hopefully helps some, but i think our logic here is unfortunately
fairly complicated which is reflected in the code comments. feel free to
suggest alternate wording or even just close this if you think our
comments currently in main are good enough
this fixes https://github.com/certbot/certbot/issues/10176 and fixes
https://github.com/certbot/certbot/issues/10257. it is based on
https://github.com/certbot/certbot/pull/10017 and ohemorange said it's
fine for me to cherry-pick their commit here
this change accomplishes a few things:
* because PYTHONPATH is no longer set to
`"$SNAP/lib/python3.12/site-packages:${PYTHONPATH}"` which evaluates to
`"$SNAP/lib/python3.12/site-packages:"` if PYTHONPATH wasn't previously
set, Certbot no longer searches for Python modules in the current
working directory which was causing #10176. i was able to reproduce this
problem with our currently released snap and verify that this change
fixes that problem
* since we no longer set PYTHONPATH at all, it won't be set in user
hooks which was causing https://github.com/certbot/certbot/issues/10257
* as an added bonus, scripts that start with `#!/usr/bin/env
/snap/certbot/current/bin/python3` as suggested
[here](https://github.com/certbot/certbot/issues/10257#issuecomment-2809421320)
are still able to find and import certbot and its dependencies so those
scripts should continue to work. i verified this is the case with manual
testing
finally, i created two news fragments here based on the text at
https://towncrier.readthedocs.io/en/stable/tutorial.html#creating-news-fragments
which says
> You can associate multiple issue numbers with a news fragment by
giving them the same contents.
when run on this PR `towncrier --draft` outputs:
```
Loading template...
Finding news fragments...
Rendering news fragments...
Draft only -- nothing has been written.
What is seen below is what would be written.
## 4.2.0.dev0 - 2025-07-31
### Changed
- Catches and ignores errors during the directory fetch for ARI checking so
that these errors do not hinder the actual certificate issuance.
([#10342](https://github.com/certbot/certbot/issues/10342))
- Removed the dependency on `pytz`.
([#10350](https://github.com/certbot/certbot/issues/10350))
### Fixed
- The Certbot snap no longer sets the environment variable PYTHONPATH stopping
it from picking up Python files in the current directory and polluting the
environment for Certbot hooks written in Python.
([#10176](https://github.com/certbot/certbot/issues/10176),
[#10257](https://github.com/certbot/certbot/issues/10257))
- Previously, we claimed to set FAILED_DOMAINS and RENEWED_DOMAINS env
variables for use by post-hooks when certificate renewals fail, but we were
not actually setting them. Now, we are.
([#10259](https://github.com/certbot/certbot/issues/10259))
- Added `--eab-hmac-alg` parameter to support custom HMAC algorithm for
External Account Binding.
([#10281](https://github.com/certbot/certbot/issues/10281))
- Certbot now always uses the server value from the renewal configuration file
for ARI checks instead of the server value from the current invocation of
Certbot. This helps prevent ARI requests from going to the wrong server if
the user changes CAs.
([#10339](https://github.com/certbot/certbot/issues/10339))
```
---------
Co-authored-by: Erica Portnoy <erica@eff.org>
fixes https://github.com/certbot/certbot/issues/10336 and fixes
https://github.com/certbot/certbot/issues/10357 using the plan at
https://github.com/certbot/certbot/issues/10336#issuecomment-3109192677
while this PR makes the renewal_time function slightly less nice, i
think us catching and handling the exceptions in certbot makes the most
sense so we can do exactly what we want around terminal and file logging
with this change, a output from a failed `sudo certbot renew` run looks
like
```
$ sudo certbot renew
Saving debug log to /var/log/letsencrypt/letsencrypt.log
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Processing /etc/letsencrypt/renewal/example.org.conf
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
An error occurred requesting ACME Renewal Information (ARI). If this problem persists and you think it's a bug in Certbot, please open an issue at https://github.com/certbot/certbot/issues/new/choose.
Certificate not yet due for renewal
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
The following certificates are not due for renewal yet:
/etc/letsencrypt/live/example.org/fullchain.pem expires on 2025-10-23 (skipped)
No renewals were attempted.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
```
`sudo certbot renew -q` produces no output and the relevant messages in
the log file look like:
```
2025-07-30 19:51:13,267:WARNING:certbot._internal.renewal:An error occurred requesting ACME Renewal Information (ARI). If this problem persists and you think it's a bug in Certbot, please open an issue at https://github.com/certbot/certbot/issues/new/choose.
2025-07-30 19:51:13,267:DEBUG:certbot._internal.renewal:Error while requesting ARI was:
Traceback (most recent call last):
File "/home/brad/certbot/acme/src/acme/client.py", line 366, in renewal_time
raise ValueError('im some error')
ValueError: im some error
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/home/brad/certbot/certbot/src/certbot/_internal/renewal.py", line 351, in _ari_renewal_time
return acme.renewal_time(cert_pem)[0]
File "/home/brad/certbot/acme/src/acme/client.py", line 370, in renewal_time
raise errors.ARIError(error_msg, now + default_retry_after) from e
acme.errors.ARIError: ('failed to fetch renewal_info URL https://acme-staging-v02.api.letsencrypt.org/acme/renewal-info/oXQaBm1Qt4YtSizBfrSNiElszRY.LMjTHFS4HPbSRMOzLrA9OZId', datetime.datetime(2025, 7, 31, 1, 51, 13, 267088))
```
something weird happened to the changelog in
https://github.com/certbot/certbot/pull/10319. a 4.2.0 entry was added
below the entry for `5.0.0 - main` despite 4.2.0 not having been
released. since it's sounding like we're expecting our next release to
be 4.2.0 and not 5.0, i merged these two changelog entries into one for
4.2.0
i also modified our setup.py files to use 4.2.0.dev0 instead of
5.0.0.dev0 altho this isn't strictly necessary because our release
script will automatically set all version numbers to whatever version we
give it on the command line before building the release
my desire to do this came from the discussion at
https://github.com/certbot/certbot/pull/10358#discussion_r2198273164
the code i'm deleting here came from
https://github.com/certbot/certbot/pull/4733
i think doing string parsing on the exception like this is convoluted
and overkill. i also agree with erica from the linked thread above that
we shouldn't be raising a ValueError here, especially when the docstring
for this function says `:raises requests.exceptions.RequestException: in
case of any problems` and doesn't mention ValueError
i prefer we do the simple thing and just delete the code. in the my
opinion unlikely event we decide polishing this important, i think we
can reconsider more complex approaches here
certbot's standalone code contains confusing references to things like
`SSLSocket` which we were hoping to deprecate in
https://github.com/certbot/certbot/issues/10284. are they relevant?
they're sure not!
certbot's standalone plugin only supports HTTP-01 so comments about
things like `ACMETLSServer` and the completely unused `certs` variable
can be deleted
furthermore, the type of the different variables named things like
`http_01_resources` were wrong in multiple places. as can be seen in
certbot's standalone code, the type is
`Set[acme_standalone.HTTP01RequestHandler.HTTP01Resource]`. this is also
[the type used in acme.standalone's
tests](https://github.com/certbot/certbot/blob/723fe64d4dbc778f1e8fd0b93320c1a30dc6fa95/acme/src/acme/_internal/tests/standalone_test.py#L78-L81)
despite the file's type annotations saying it takes a different type. i
think the incorrect type annotations were never caught because mypy
can't fully make sense of our overly complex server classes here
finally, `from __future__ import annotations` was added to make [forward
references in type
annotations](https://mypy.readthedocs.io/en/stable/cheat_sheet_py3.html#forward-references)
easier
fixes https://github.com/certbot/certbot/issues/10308
my thinking here was if the spec forbids checking ARI for expired certs,
this check should happen directly in the renewal_time function. if we do
that, what's its most useful response? error? return None? return a
datetime in the past?
i feel the latter is most helpful. tell the caller to renew now rather
than erroring out or giving it no suggestion about when it should renew
it probably doesn't matter much, but i think this would be nice to have
for 4.1.0 as it fixes a (minor) spec compliance issue in our ARI
implementation that is being released
this is a follow up from https://github.com/certbot/certbot/pull/10286
and related to https://github.com/certbot/certbot/issues/10302
sorry i initially missed this! in #10286 our tests were just yelling at
me about the different augeas package needed, but python headers and a
compiler are also needed for things to work with an updated version of
python-augeas
i don't believe we need this change in our macOS instructions because:
1. homebrew doesn't split up python packages the way many linux distros
do. there is no equivalent python-dev package
2. if you're using homebrew, you already have a compiler because
[homebrew requires command line tools for
xcode](https://docs.brew.sh/Installation#macos-requirements)
3. "it works on my machine"
it looks like https://github.com/certbot/certbot/pull/10098 introduced a
couple bugs into this file:
1.
[RSAPrivateKeys](https://cryptography.io/en/latest/hazmat/primitives/asymmetric/rsa/#cryptography.hazmat.primitives.asymmetric.rsa.RSAPrivateKey)
don't have a `public_bytes` method
2. `cryptography.x509` wasn't imported and
[load_pem_x509_certificate](https://cryptography.io/en/latest/x509/reference/#cryptography.x509.load_pem_x509_certificate)
takes bytes, not a string
i think avoiding this is unfortunately difficult as this file has no
tests, but it was useful for me just now when testing
https://github.com/certbot/certbot/pull/10283 so i wanted to fix it up
i also changed the script to initially create the account without an
email address as the fake@example.com email causes registration with
LE's staging server to fail early in execution
with the changes in this PR changes, if you:
1. change the value of
[DOMAIN](https://github.com/certbot/certbot/blob/0075104805e2e4fd9cf0e6ec69d80ab203e03655/acme/examples/http01_example.py#L57)
to a domain pointing at your machine
2. as root, activate your certbot dev environment, and run `python
acme/examples/http01_example.py `
it will fail late in the script with:
```
Traceback (most recent call last):
File "/home/brad/certbot/acme/examples/http01_example.py", line 237, in <module>
example_http()
~~~~~~~~~~~~^^
File "/home/brad/certbot/acme/examples/http01_example.py", line 223, in example_http
regr = client_acme.update_registration(
regr.update(
...<3 lines>...
)
)
File "/home/brad/certbot/acme/src/acme/client.py", line 101, in update_registration
updated_regr = self._send_recv_regr(regr, body=body)
File "/home/brad/certbot/acme/src/acme/client.py", line 373, in _send_recv_regr
response = self._post(regr.uri, body)
File "/home/brad/certbot/acme/src/acme/client.py", line 392, in _post
return self.net.post(*args, **kwargs)
~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
File "/home/brad/certbot/acme/src/acme/client.py", line 766, in post
return self._post_once(*args, **kwargs)
~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
File "/home/brad/certbot/acme/src/acme/client.py", line 781, in _post_once
response = self._check_response(response, content_type=content_type)
File "/home/brad/certbot/acme/src/acme/client.py", line 630, in _check_response
raise messages.Error.from_json(jobj)
acme.messages.Error: urn:ietf:params:acme:error:invalidContact :: The provided contact URI was invalid :: Unable to update account :: invalid contact: contact email has forbidden domain "example.org"
```
if you also change [this email
variable](https://github.com/certbot/certbot/blob/0075104805e2e4fd9cf0e6ec69d80ab203e03655/acme/examples/http01_example.py#L223)
to a valid email address, the script will run successfully
i wanted this for testing
https://github.com/certbot/certbot/issues/10190
alex started working on this in
https://github.com/certbot/certbot/pull/9207 years ago, but pebble
didn't end up doing a release containing his work while he was still
regularly contributing to certbot. this has now changed though
before this PR, our integration tests only worked on amd64 linux
systems. with this PR, i've successfully run our integration tests on
all combinations of the architectures amd64 and arm64 and the OSes linux
and macos
---------
Co-authored-by: ohemorange <erica@eff.org>
in https://github.com/certbot/certbot/pull/9124 we had the problem of
certbot-nginx's `Addr.fromstring` method possibly returning None which
is not possible in the `Addr` method in the certbot base class or in
certbot-apache. we fixed this by telling mypy the common
`Addr.fromstring` method returns an `Optional[Addr]` (despite it
actually always returning an `Addr`) and then unnecessarily complicating
certbot-apache's code a bit. the need for extra complexity with this
approach is going even further in
https://github.com/certbot/certbot/pull/10151 where we have to use
`cast` to assure mypy that the type isn't actually `Optional`. i
personally don't like all this
fixes https://github.com/certbot/certbot/issues/10131
this seems simple enough, but i also requested alex's review as a quick
sanity check if he doesn't mind providing one
i've verified this fixes the problem and that PKCS#8 was used in certbot
3.0.1
fixes#10105
this PR updates our minimally required cryptography and pyopenssl
versions as well as updating our policy for choosing minimum dependency
versions generally
before this PR, we were trying to keep compatibility with packages
available in EPEL 9 using the python 3 version available in RHEL 9.
after the discussion in #10105 we decided not to do this anymore
because:
* EPEL 9 may not want to update to certbot 3.0+ anyway because of our
backwards incompatible changes from certbot 2.x
* RHEL 9 appstream repos now contain newer versions of many of our
dependencies for newer versions of python
* alternate installation methods for RHEL 9 based users including our
snaps and pip are available
on a call we then discussed what distro repositories we should track
instead of EPEL 9. our docs previously said Debian sid/unstable, but we
felt this as unnecessary because Debian sid can and does change very
quickly. if we wanted a new dependency there, Debian could probably
accommodate it
we also considered RHEL 10 + EPEL 10, however, these repos are not even
stable yet and certbot and many of its dependencies are not yet packaged
there at all
for these reasons, plus many of the reasons we decided to upgrade past
EPEL 9 with the default python 3 version there, we decided that at least
for now, we will remove any linux distro considerations when choosing
minimal dependency versions of certbot
as i wrote in the contributing docs, we may choose to reconsider this
plan if there are requests for us to do so, but based on the information
above, we are not sure this will ever even happen and removing this
constraint significantly simplifies development of certbot
this hopefully at least helps the problem hit at
https://github.com/certbot/certbot/pull/10126#discussion_r1909714276
i took this approach because in my experience, linux specific shell
commands have crept into our scripts repeatedly over the years so i
think just having macOS devs use the linux versions is much more
reliable. it's what i've personally been doing for years now
when reviewing https://github.com/certbot/certbot/pull/10126 and running
`tools/pinning/oldest/repin.sh` using a freshly created dev environment,
i was repeatedly given the message
> The "poetry.dev-dependencies" section is deprecated and will be
removed in a future version. Use "poetry.group.dev.dependencies"
instead.
i believe this section was generated automatically by poetry's tooling
when it created the initial boilerplate file for us, but we don't use
it, so i just deleted the section which makes the warnings disappear
this PR hopefully improves two things that i hit while working on #10035
1) i found that repinning our dependencies took ~6 minutes!
digging into it a bit, the biggest culprit i found was the inclusion of
`--no-cache-dir` here which seemed to cause poetry to redownload the
same packages over and over in a single run. this comes from
https://github.com/certbot/certbot/pull/9453 which fixed a problem i
(but not alex) was having with a major performance penalty. i removed
the flag here and instead included instructions on clearing poetry's
caches in case anyone ever hits this in the future. with this change,
the script now takes about 40 seconds on my laptop
2) every run of this script ended with the output:
```
Warning: poetry-plugin-export will not be installed by default in a
future version of Poetry.
In order to avoid a breaking change and make your automation
forward-compatible, please install poetry-plugin-export.
explicitly. See https://python-poetry.org/docs/plugins/#using-plugins
for details on how to install a plugin.
To disable this warning run 'poetry config warnings.export false'.
```
setting `POETRY_WARNINGS_EXPORT=false` fixes this which i believe is
safe to do because of
https://github.com/certbot/certbot/blob/2c8609464c3e371053f78e515fc7beb26446f5a3/certbot/setup.py#L53-L56
i hit this when working on https://github.com/certbot/certbot/pull/10076
where i found that updating all our dependencies no longer worked
because of new deprecations in pyopenssl. this pr fixes that
This is another and very minor piece of https://github.com/certbot/certbot/issues/9988.
We've done nothing to warn/migrate installations using the old `certbot-route53:auth` plugin name and installations like that still exist according to https://gist.github.com/bmw/aceb69020dceee50ba827ec17b22e08a. We could try to warn/migrate these users for a future release or decide it's niche enough that we'll just let it break, but I think it's easy enough to keep the simple shim around.
This PR just moves the code raising a deprecation warning into `_internal` as part of cleaning up all deprecation warnings I found in https://github.com/certbot/certbot/issues/9988. I manually tested this with a Certbot config using the `certbot-route53:auth` plugin name and renewal worked just fine.
The `gnupg` package from Homebrew only installs a `gpg` binary, not a `gpg2` binary. I had previously worked around this by manually creating an alias, but I think we can do better.
GPG version 1 is ancient and [hasn't seen a release since 2006](https://gnupg.org/download/release_notes.html). Additionally, `gpg` has referred to GPG 2 in Ubuntu since at least 20.04 which is the oldest non-EOL'd version as of writing this so I think this change is safe to make.
while working on https://github.com/certbot/certbot/issues/9938, i updated our dependencies which updated mypy introducing new errors that mypy wanted me to fix. i think this makes the regularly necessary process of updating our dependencies too tedious and we should instead pin our linters that do this to a specific version and update them manually as desired. we already do this with pylint in the lines above my changes in this PR for the same reason
* Set the delegated field in Lexicon config to bypass subdomain resolution (#9821)
The Lexicon-based DNS plugins use a mechanism to determine which actual segment of the input domain is actually the DNS zone in which the DNS-01 challenge has to be initiated (eg. `subdomain.domain.com` or `domain.com` for input `subdomain.domain.com`): they tries recursively to configure Lexicon and initiate authentication from the most specific to most generic domain segment, and select the first segment where Lexicon stop erroring out.
This mechanism broke with #9746 because now the plugins call Lexicon client instead of the underlying providers, and the client makes guess on the actual domain requested. Typically for `subdomain.domain.com` it will actually try to authenticate against `domain.com`, and so the mechanism above does not work anymore.
This PR fixes the issue by using the `delegated` field in Lexicon config each time the plugin needs it. This field is designed for this kind of purpose: it will instruct Lexicon what is the actual DNS zone domain instead of guessing it.
I tested the change with one of my OVH account. The expected behavior is re-established and the plugin is able to test `subdomain.domain.com` then `domain.com` as before.
Fixes#9791Fixes#9818
(cherry picked from commit cf4f07d17e)
* add changelog entry for 9821 (#9822)
(cherry picked from commit 7bb85f8440)
---------
Co-authored-by: Adrien Ferrand <adferrand@users.noreply.github.com>
* helpful: fix handling of abbreviated ConfigArgparse arguments (#9796)
* helpful: fix handling of abbreviated ConfigArgparse arguments
ConfigArgparse allows for "abbreviated" arguments, i.e. just the prefix
of an argument, but it doesn't set the argument sources in these cases.
This commit checks for those cases and sets the sources appropriately.
* failing to find an action raises an error instead of logging
* Update changelog
* Add handling for short arguments, fix equals sign handling
These were silently being dropped before, possibly leading to instances
of `NamespaceConfig.set_by_user()` returning false negatives.
(cherry picked from commit 11e17ef77b)
* Fix finish_release.py (#9800)
* response is value
* rename vars
(cherry picked from commit a96fb4b6ce)
* Merge pull request #9762 from certbot/docs/yaml-config
Add YAML files for Readthedocs requirements
(cherry picked from commit 44046c70c3)
* Update Lexicon requirements to stabilize certbot-dns-ovh behavior (#9802)
* Update minimum Lexicon version required for certbot-dns-ovh
* Add types
* FIx mypy
* Fix lint
* Fix BOTH lint and mypy
(cherry picked from commit 5cf5f36f19)
* simplify code (#9807)
(cherry picked from commit 6f7b5ab1cd)
* Include linting fixes from 8a95c03
---------
Co-authored-by: Will Greenberg <willg@eff.org>
Co-authored-by: Alexis <alexis@eff.org>
Co-authored-by: Adrien Ferrand <adferrand@users.noreply.github.com>
* update quickstart and remove os import
* simplify theme use
* list sphinx_rtd_theme as extension
Our docs builds failed last night, presumably because #9754 updated `sphinx_rtd_theme` which changed some unknown thing.
Looking into it, our usage of this project was very unconventional. Following the code comment I deleted in this PR to https://docs.readthedocs.io/en/stable/faq.html#i-want-to-use-the-read-the-docs-theme-locally, simple instructions are given to put the following in your `conf.py` file:
```
extensions = [
...
'sphinx_rtd_theme',
]
html_theme = "sphinx_rtd_theme"
```
I did this instead of the more complicated logic we were using and all builds passed locally. I also triggered a build on readthedocs with these changes which also passed.
