So merging the study from @bmw and me, here is what happened.
Each invocation of `certbot.logger.post_arg_parse_setup` create a file handler on `letsencrypt.log`. This function also set an atexit handler invoking `logger.shutdown()`, that have the effect to close all logger file handler not already closed at this point. This method is supposed to be called when a python process is close to exit, because it makes all logger unable to write new logs on any handler.
Before #6667 and this PR, for tests, the atexit handle would be triggered only at the end of the pytest process. It means that each test that launches `certbot.logger.post_arg_parse_setup` add a new file handler. These tests were typically connecting the file handler on a `letsencrypt.log` located in a temporary directory, and this directory and content was wipped out at each test tearDown. As a consequence, the file handles, not cleared from the logger, were accumulating in the logger, with all of them connected to a deleted file log, except the last one that was just created by the current test. Considering the number of tests concerned, there were ~300 file handler at the end of pytest execution.
One can see that, on prior #6667, by calling `print(logger.getLogger().handlers` on the `tearDown` of these tests, and see the array growing at each test execution.
Even if this represent a memory leak, this situation was not really a problem on Linux: because a file can be deleted before it is closed, it was only meaning that a given invocation of `logger.debug` for instance, during the tests, was written in 300 log files. The overhead is negligeable. On Windows however, the file handlers were failing because you cannot delete a file before it is closed.
It was one of the reason for #6667, that added a call to `logging.shutdown()` at each test tearDown, with the consequence to close all file handlers. At this point, Linux is not happy anymore. Any call to `logger.warn` will generate an error for each closed file handler. As a file handler is added for each test, the number of errors grows on each test, following an arithmetical suite divergence.
On `test_sdists.py`, that is using the bare setuptools test suite without output capturing, we can see the damages. The total output takes 216000 lines, and 23000 errors are generated. A decent machine can support this load, but a not a small AWS instance, that is crashing during the execution. Even with pytest, the captured output and the memory leak become so large that segfaults are generated.
On the current PR, the problem is solved, by resetting the file handlers array on the logging system on each test tearDown. So each fileHandler is properly closed, and removed from the stack. They do not participate anymore in the logging system, and can be garbage collected. Then we stay on always one file handler opened at any time, and tests can succeed on AWS instances.
For the record, here is all the places where the logging system is called and fail if there is still file handlers closed but not cleaned (extracted from the original huge output before correction):
```
Logged from file account.py, line 116
Logged from file account.py, line 178
Logged from file client.py, line 166
Logged from file client.py, line 295
Logged from file client.py, line 415
Logged from file client.py, line 422
Logged from file client.py, line 480
Logged from file client.py, line 503
Logged from file client.py, line 540
Logged from file client.py, line 601
Logged from file client.py, line 622
Logged from file client.py, line 750
Logged from file cli.py, line 220
Logged from file cli.py, line 226
Logged from file crypto_util.py, line 101
Logged from file crypto_util.py, line 127
Logged from file crypto_util.py, line 147
Logged from file crypto_util.py, line 261
Logged from file crypto_util.py, line 283
Logged from file crypto_util.py, line 307
Logged from file crypto_util.py, line 336
Logged from file disco.py, line 116
Logged from file disco.py, line 124
Logged from file disco.py, line 134
Logged from file disco.py, line 138
Logged from file disco.py, line 141
Logged from file dns_common_lexicon.py, line 45
Logged from file dns_common_lexicon.py, line 61
Logged from file dns_common_lexicon.py, line 67
Logged from file dns_common.py, line 316
Logged from file dns_common.py, line 64
Logged from file eff.py, line 60
Logged from file eff.py, line 73
Logged from file error_handler.py, line 105
Logged from file error_handler.py, line 110
Logged from file error_handler.py, line 87
Logged from file hooks.py, line 248
Logged from file main.py, line 1071
Logged from file main.py, line 1075
Logged from file main.py, line 1189
Logged from file ops.py, line 122
Logged from file ops.py, line 325
Logged from file ops.py, line 338
Logged from file reporter.py, line 55
Logged from file selection.py, line 110
Logged from file selection.py, line 118
Logged from file selection.py, line 123
Logged from file selection.py, line 176
Logged from file selection.py, line 231
Logged from file selection.py, line 310
Logged from file selection.py, line 66
Logged from file standalone.py, line 101
Logged from file standalone.py, line 88
Logged from file standalone.py, line 97
Logged from file standalone.py, line 98
Logged from file storage.py, line 52
Logged from file storage.py, line 59
Logged from file storage.py, line 75
Logged from file util.py, line 56
Logged from file webroot.py, line 165
Logged from file webroot.py, line 186
Logged from file webroot.py, line 187
Logged from file webroot.py, line 204
Logged from file webroot.py, line 223
Logged from file webroot.py, line 234
Logged from file webroot.py, line 235
Logged from file webroot.py, line 237
Logged from file webroot.py, line 91
```
* Reapply #6667
* Make setuptools delegates tests execution to pytest, like in acme module.
* Clean handlers at each tearDown to avoid memory leaks.
* Update changelog
I observed that the current set of oldest requirements do not correspond to any environment, except the specific Xenial image in Travis CI (and standard Xenial containers will also fail).
It is because the requirements make cryptography and requests fail against standard libraries available in the typical Linux distributions that are targeted by the oldest requirements approach (Centos 6, Centos 7, Xenial, Jessie).
This PR fixes that, by aligning the minimal version requirements of cryptography and requests to the maximal versions that are available on Centos 6. Centos 7, Jessie and Xenial stay unusable with oldest requirements for other reasons, but at least one old and supported Linux distribution is able to run the tests with oldest requirements out of the box.
A test is also corrected to match the expected error message that old versions of urllib3 will raise.
* Rough draft of External Account Binding.
* Remove parameter --eab and namespace kid and hmac. Also add parameters to "register" subcommand.
* Refactor as much as possible of the EAB functionality into ExternalAccountBinding class.
* Remove debug line.
* Added external account binding to Directory.Meta.
* Rename to account_public_key, hmac_key and make some non-optional.
Rename command line argument to --eab-hmac-key.
* Error out when the server requires External Account Binding and the user
has not supplied kid and hmac key.
* Remove whitespace.
* Refactor a bit to make it possible to set the url argument.
* Move from_data method into client.
* Revert "Move from_data method into client."
This reverts commit 8963fae
* Refactored to use json field on Registration.
* Inherit from object according to Google Python Style Guide.
* Move to two separate ifs.
* Get tests to pass after External Account Binding additions.
* messages.py back to 100% test coverage with some EAB tests.
* .encode() this JSON key.
* Set eab parameter default values to None.
* * Remove unnecessary public key mock on most of the test.
* Restructure the directory mock to be able to mock both True and False for externalAccountRequired easily.
* Add EAB client tests.
* Move external_account_required check into BackwardsCompatibleClientV2 to be able to mock it.
* Update versions.
* Try 0.29.0.
* Revert "Try 0.29.0."
This reverts commit 5779509
* Try 0.29.0 again.
* Try this.
* Fix pylint failures.
* Add tests for external_account_required method.
* Test not needed, avoid:
************* Module acme.client_test
C: 1, 0: Too many lines in module (1258/1250) (too-many-lines)
* Move real external_account_required method into ClientV2 and pass through to it in BackwardsCompatibleClientV2.
* Handle missing meta key in server ACME directory.
* Add docstring for BackwardsCompatibleClientV2.external_account_required().
* Add tests for BackwardsCompatibleClientV2.external_account_required().
* Fix coverage for ACMEv1 code in BackwardsCompatibleClientV2.
* Disable pylint too-many-lines check for client_test.py.
* Fix versions.
* Remove whitespace that accidently snuck into an earlier commit.
* Remove these two stray whitespaces also.
* And the last couple of whitespaces.
* Add External Account Binding to changelog.
* Add dev0 suffix to setup.py.
Co-Authored-By: robaman <robert@kastel.se>
* Set to "-e acme[dev]" again.
Co-Authored-By: robaman <robert@kastel.se>
Fixes#4686.
In Sphinx 1.6, they changed how they handle images in latex and PDF files. You can learn more about this by reading the linked issue (or I can answer any questions), but the shortish version is we now need to use the extension sphinx.ext.imgconverter. This is only available in Sphinx 1.6+.
I also updated our pinned versions to use the latest Sphinx and a new dependency it pulled in called sphinxcontrib-websupport. To build the latex and PDF docs, you must first run:
apt-get install imagemagick latexmk texlive texlive-latex-extra
Afterwards, if you create the normal Certbot dev environment using this branch, activate the virtual environment, and from the root of the repo run make -C docs clean latex latexpdf, you'll successfully build the PDF docs.
* fix#4686
* bump minimum Sphinx req
* Use greater than or equal to in requirements.
This changes the existing requirements using strictly greater than to greater
than or equal to so that they're more conventional.
* Use >= for certbot-postfix.
Despite it previously saying 'certbot>0.23.0', certbot-postfix/local-oldest-requirements.txt was pinned to 0.23.0 so let's just use certbot>=0.23.0.
* Remove unneeded sys import.
Once upon a time we needed this in some of these setup.py files because we were
using sys in the file, but we aren't anymore so let's remove the import.
* use setuptools instead of distutils
* check_untyped_defs in mypy with clean output for acme
* test entire acme module
* Add typing as a dependency because it's only in the stdlib for 3.5+
* Add str_utils, modified for python2.7 compatibility
* make mypy happy in acme
* typing is needed in prod
* we actually only need typing in acme so far
* add tests and more docs for str_utils
* pragma no cover
* add magic_typing
* s/from typing/from magic_typing/g
* move typing to dev_extras
* correctly set up imports
* remove str_utils
* only type: ignore for OpenSSL.SSL, not crypto
* Since we only run mypy with python3 anyway and we're fine importing it when it's not actually there, there's no actual need for typing to be present as a dependency
* comment magic_typing.py
* disable wildcard-import im magic_typing
* disable pylint errors
* add magic_typing_test
* make magic_typing tests work alongside other tests
* make sure temp_typing is set
* add typing as a dev dependency for python3.4
* run mypy with python3.4 on travis to get a little more testing with different environments
* don't stick typing into sys.modules
* reorder imports
Fixes#5490.
There's a lot of possibilities discussed in #5490, but I'll try and explain what I actually did here as succinctly as I can. Unfortunately, there's a fair bit to explain. My goal was to break lockstep and give us tests to ensure the minimum specified versions are correct without taking the time now to refactor our whole test setup.
To handle specifying each package's minimum acme/certbot version, I added a requirements file to each package. This won't actually be included in the shipped package (because it's not in the MANIFEST).
After creating these files and modifying tools/pip_install.sh to use them, I created a separate tox env for most packages (I kept the DNS plugins together for convenience). The reason this is necessary is because we currently use a single environment for each plugin, but if we used this approach for these tests we'd hit issues due to different installed plugins requiring different versions of acme/certbot. There's a lot more discussion about this in #5490 if you're interested in this piece. I unfortunately wasted a lot of time trying to remove the boilerplate this approach causes in tox.ini, but to do this I think we need negations described at complex factor conditions which hasn't made it into a tox release yet.
The biggest missing piece here is how to make sure the oldest versions that are currently pinned to master get updated. Currently, they'll stay pinned that way without manual intervention and won't be properly testing the oldest version. I think we should solve this during the larger test/repo refactoring after the release because the tests are using the correct values now and I don't see a simple way around the problem.
Once this lands, I'm planning on updating the test-everything tests to do integration tests with the "oldest" versions here.
* break lockstep between packages
* Use per package requirements files
* add local oldest requirements files
* update tox.ini
* work with dev0 versions
* Install requirements in separate step.
* don't error when we don't have requirements
* install latest packages in editable mode
* Update .travis.yml
* Add reminder comments
* move dev to requirements
* request acme[dev]
* Update pip_install documentation
* Drop support for EOL Python 2.6
* Use more helpful assertIn/NotIn instead of assertTrue/False
* Drop support for EOL Python 3.3
* Remove redundant Python 3.3 code
* Restore code for RHEL 6 and virtualenv for Py2.7
* Revert pipstrap.py to upstream
* Merge py26_packages and non_py26_packages into all_packages
* Revert changes to *-auto in root
* Update by calling letsencrypt-auto-source/build.py
* Revert permissions for pipstrap.py
* Use pipstrap to install a good version of pip
* Use pytest in cb-auto tests
* Remove nose usage in auto_test.py
* remove nose dev dep
* use pytest in test_tests
* Use pytest in tox
* Update dev dependency pinnings
* remove nose multiprocess lines
* Use pytest for coverage
* Use older py and pytest for old python versions
* Add test for Error.__str__
* pin pytest in oldest test
* Fix tests for DNS-DO plugin on py26
* Work around bug for Python 3.3
* Clarify dockerfile comments
* cert signature validation for certificates subcommand + a test
* refactoring validation + adding in a check for making sure that the private key matches the certificate
* adding testing certs
* assertIsNone(x) -> assertEqual(None,x) to unbreak the py2.6 tests
* modifying test_verifu_renewable_cert_failure to hopefully appease python 3 test timeouts
* updating cryptography to be >=1.2 so that we can use verify
* removing unused, old testing certificate
* adding better error handling/logging
* adding test for IOError
* switching to a 2048 bit rsa key
* Mention python 3 support in setup.py
* Build universal (py2 and py3 compatible) wheels
* Mention Python 3.3+ support in docs
* we work on python 3.6 too
* add fasteners as a dependency
* add LOCK_FILE constant
* Add lock file to Certbot
* Move code to _run_subcommand
* move lock file path into CLI_CONSTANTS
* add --lock-path flag
* move locking code to separate function
* Add TestAcquireFileLock
* assert we log
* test lock contention
* add fasteners to certbot-auto
* Use a different lock file for each test in MainTest
* stop conditionally pinning mock version in acme
* stop conditionally pinning mock version in certbot
* stop conditionally pinning mock version in apache
* stop conditionally pinning mock version in nginx
* stop conditionally pinning mock version in letshelp
* stop conditionally pinning mock version in compatibility-test
The primary motivation is to avoid a branch, giving bugs one fewer place to hide. But, as a bonus, more people get a more bugfixed version of argparse. (To use the example from the argparse docs, people stuck on Python 3.2.3 can get bugfixes that made it into the stdlib only in 3.2.4.)
These changes allow developers to run tests directly from the host machine using Docker, and to enable ipdb inside the container.
docker-compose.yml is upgraded to version 2 format. This means that you need docker-engine version >= 1.10.0 instead of previous requirement of version >= 1.9.1. The reason for this is to be able to use custom Dockerfile (Dockerfile-dev in this case) in build context.
ipdb has been added to dev dependencies to be able to be able to debug the code without installing it on every docker run. This is also what we recommend for debugging in the developer documentation, so there really is no reason not to install it with the dev dependencies.
setuptools is being upgraded to a newer version to be able to run coverage tests. This was using the older version of setuptools for some reason, and without the upgrade, coverage tests would fail horribly. Upgrading remedies the situation.
Few examples:
Run unit tests for certbot-apache
`docker-compose run --rm --service-ports development bash -c 'cd src;nosetests -v certbot-apache'`
Run coverage tests
`docker-compose run --rm --service-ports development bash -c 'cd src;./tox.cover.sh'`
Run linter
`docker-compose run --rm --service-ports development bash -c 'cd src;tox -e lint'`
* Start of combined manual/script plugin
* Return str from hooks.execute, not bytes
* finish manual/script rewrite
* delete old manual and script plugins
* manually specify we want chall.token
* use consistent quotes
* specify chall for uri
* s/script/hook
* fix spacing on instructions
* remove unneeded response argument
* make achall more helpful
* simplify perform
* remove old test files
* add start of manual_tests
* fix ParseTest.test_help
* stop using manual_test_mode in cli tests
* Revert "make achall more helpful"
This reverts commit 54b01cea30.
* use bad response/validation methods on achalls
* simplify perform and cleanup environment
* finish manual tests
* Add HTTP manual hook integration test
* add manual http scripts
* Add manual DNS script integration test
* remove references to the script plugin
* they're hooks, not scripts
* add --manual-public-ip-logging-ok to integration tests
* use --pref-chall for dns integration
* does dns work?
* validate hooks
* test hook validation
* Revert "does dns work?"
This reverts commit 1224cc2961.
* busy wait in manual-http-auth
* remove DNS script test for now
* Fix challenge prefix and add trailing .
* Add comment about universal_newlines
* Fix typo from 0464ba2c4
* fix nits and typos
* Generalize HookCOmmandNotFound error
* Add verify_exe_exists
* Don't duplicate code in hooks.py
* Revert changes to hooks.py
* Use consistent hook error messages
* Script plugin initial commit
* Fix auth script path
* Return correct responses
* Added DNS-01 support
* Report the challenge pref correctly
* Use config root from certbot constants rather than hardcoded
* Remove prehook and rename posthook to cleanup for clarity
* Refactoring
* Docs
* Refactoring
* Refactoring continued, working now
* Use global preferred-challenges argument in favor of local
* Added http-01 as fallback challenge if not defined
* Do not continue if auth script not defined
* Skip unnecessary steps when running
* Read config values from correct places
* Tests and minor fixes
* Make Python 2.6 happy again
* Added CERTBOT_AUTH_OUTPUT and better tests
* Lint & Py3 fixes
* Make Python 2.6 happy again
* Doc changes
* Refactor hook execute and reuse in script plugin
* Refactored hook validation
* Added long_description for plugin help text
* Refactored env var writing
