* repin current
* repin oldest
* csr must have version set to zero
* only set PIP_USE_PEP517 for macOS
* experiment with brew update git failure workaround
* letstest: -ubuntu18.04 +centos9stream +debian11
* letstest: username for centos 9 stream is ec2-user
This is mentioned on https://centos.org/download/aws-images/
* ensure mod_ssl is installed
in centos 9 stream, apache has to be restarted after mod_ssl is
installed, or the snakeoil certificates will not be present and
apache won't start.
this also removes nghttp2 being installed as the relevant bug
is long fixed.
* dns-rfc2136: add test coverage for PR #9672
* fix compatibility with oldest dnspython
* rename test to be more descriptive
Co-authored-by: ohemorange <ebportnoy@gmail.com>
---------
Co-authored-by: ohemorange <ebportnoy@gmail.com>
* remove pointless paragraph about --server and wildcards
* docs: update help text for --dry-run and --staging
* docs: update "Changing the ACME Server" for --dry-run
* add note about webserver reloads
In #9127, where @osirisinferi added the `show_account` verb, I made a call not to include the thumbprint in the output of `certbot show_account`.
In hindsight, and after a community member asked for this feature, I think it's better to include it.
It is useful on occasion and `show_account` is fairly specialized anyway. It's only really good for getting your account URL for rate limit increases, checking your contacts, and (now) and doing *magic* with the thumbprint for stateless/distributed HTTP-01 responders.
Without this feature, a clever user might figure out their thumbprint by doing a `certonly --manual --preferred-challenges http` request, but most users would probably be lost.
* show_account: display account thumbprint
* use local key for display
* certbot-ci: boulder will now only supports port 80 for http-01
* forgot to actually use the http_01_port argument
* print the port the proxy listens on
* try allow binding to privileged ports
Based on my design [here](https://docs.google.com/document/d/1jGh_bZPnrhi96KzuIcyCJfnudl4m3pRPGkiK4fTo8e4/edit?usp=sharing).
Fixes https://github.com/certbot/certbot/issues/4634 and https://github.com/certbot/certbot/issues/4635.
- [x] Deprecate `NamespaceConfig.csr_dir`,`NamespaceConfig.key_dir`, ~~`constants.CSR_DIR` and `constants.KEY_DIR`~~. (`constants` is `_internal` so we can just delete it eventually).
- [x] Update `certbot.crypto_util.generate_csr` and `.generate_key` to make `csr_dir` and `key_dir` optional, respectively.
- [x] Change `certbot._internal.client.Client.obtain_certificate` to no longer include `csr_dir` and `key_dir` to the `.generate_csr` and `.generate_key` calls, respectively.
- Automatically delete unwanted lineage items:
- [x] In `certbot._internal.storage.RenewableCert`, add a function to truncate the lineage history according to the criteria (keep the current and the 5 prior certificates).
- [x] Add a test suite for `truncate`
- [x] In `certbot._internal.renewal.renew_cert`, call the lineage truncation function after the symlinks have been updated for the renewal.
* Stop writing new files to /csr and /keys
* storage: add lineage truncation
* remove unused code
* deprecate keys_dir and csr_dir
* update CHANGELOG
* just keep 5 prior certificates, dont be clever with expiry
* docs: remove reference to /archive and /keys
* filter {csr,key}_dir deprecations directly in tests
* nginx: capitalise product names in warning message properly
* nginx: don't crash on encountering lua directives, warn instead
* add tests
* undo excess newline
* fix oldest tests: use old camelCase function name
* add missing newline in new testdata
* add tests for _by_lua, which should parse fine
At the time this section was written, it was all about the introduction of support for ECDSA and how users can start taking advantage of that support.
Now that we use ECDSA by default, this piece of documentation probably should serve a new purpose. My idea here is to document the new behavior that we have in 2.0: new key type on new certificates, old certificates will keep their existing key type.
Users may now be going in the reverse direction with their changes ("I got an ECDSA certificate but I need RSA because I have an old load balancer appliance!") so I have also updated some section titles to be less about ECDSA and more about Key Types in general.
Fixes#9442.
Fixes#7206.
I think it's about time we did this:
- `dnssec-keygen` on new distros doesn't support the HMAC algorithms anymore, so our instructions don't work.
- The oldest distros we support are Debian Buster (`9.11.5.P4+dfsg-5.1+deb10u7`) and CentOS 7 (`9.11.4-26.P2.el7_9.9`), which ship `tsig-keygen` and support `HMAC-SHA512`.
* dont superfluously ask whether to renew, when changing key type
* reorder conditions
this prevents "Certificate not yet due for renewal" being printed
* and replace superfluous mock
* mock renewal.should_renew
* acme: remove Client and BackwardsCompatibleClientV2
* remove ClientTestBase and some unused variables
* add ClientV2.get_directory
* tweak ToS callback code
* acme: update example to use ClientV2.get_directory
* simplify ToS callback further into one step
* further removal of acmev1-related code
- remove acme.client.ClientBase
- remove acme.mixins.VersionedLEACMEMixin
- remove acme.client.DER_CONTENT_TYPE
- remove various ACMEv1 special cases
- remove acme.messages.ChallengeResources.combinations
* remove .mixins.ResourceMixin, fields.resource, fields.Resource
and resource field from various .message classes.
* simplify acme.messages.Directory:
- remove Directory.register
- remove HasResourceType and GenericHasResourceType
- remove ability to look up Directory resources by anything other
than the exact field name in RFC8555 (section 9.7.5)
* remove acme.messages.OLD_ERROR_PREFIX and support the old prefix
* remove acme.mixins
* reorder imports
* add comment to Directory about resource lookups
* s/new-cert/newOrder/
* get rid of `resource` sillyness in tests
* remove acmev1 terms-of-service support from directory
* deprecate more attributes in acme
* Deprecate .Authorization.combinations by renaming the field and
deprecating in getters/setters
* Silence deprecation warnings from our own imports of acme.mixins
Co-authored-by: Brad Warren <bmw@users.noreply.github.com>
* deps: remove pyjwt dependency
* pinning: strip extras from dependencies
`poetry export` outputs in requirements.txt format, which is now
apparently producing "dep[extra]==...". We are using this output
as the constraints file for pip and pip's new resolver does not
permit extras in the constraints file.
This change filters out the extras specifiers.
* repin current dependencies
* fix new pylint complaints
* silence lint about distutils.version
We have already deprecated the function and it'll be removed in
2.0.
* docs: set sphinx language to 'en'
this is emitting a warning and failing the build
* Revert "pinning: strip extras from dependencies"
This reverts commit 11268fd231.
* pin poetry back to avoid extras issue
* repin
* fix new mypy complaints in acme/
* docs: how to override the trusted CA certificates
* Update certbot/docs/using.rst
Co-authored-by: ohemorange <ebportnoy@gmail.com>
Co-authored-by: ohemorange <ebportnoy@gmail.com>
@osirisinferi pointed out [in chat](https://opensource.eff.org/eff-open-source/pl/y5whp5ny378wuedi8gd7995qbo) that the way this entry was written, suggested that `--new-key` might affect whether `--reuse-key` is set or not.
I think the second sentence was the main culprit, so I've nixed it and replaced it with a reminder about our other flags.
This maybe calls out more for a documentation section but let's fix this quickly before the release.
* certbot-ci: fix challtestsrv address for boulder-v2
The port is no longer exposed on the Docker host.
* vary the challtestsrv URL by acme server
* fix mypy
* fix comment
Co-authored-by: ohemorange <ebportnoy@gmail.com>
Co-authored-by: ohemorange <ebportnoy@gmail.com>
* improve handling and ux of unexpected key type migration
* update unit tests
* update integration tests
* if --cert-name and --key-type are set, dont prompt
* Certificate issuing on Window while having web.confing and more then one domain in request
* add a test
* update changelog
Co-authored-by: Serghei Trufkin <Serghei.Trufkin@Technosoft.md>
* docs: describe how to modify renewal config
* Apply suggestions from code review
Co-authored-by: ohemorange <ebportnoy@gmail.com>
* reword warning about manual modifications
* explain the flags in the --force-renewal command
Co-authored-by: ohemorange <ebportnoy@gmail.com>