Verified Commit 0f5fa3a9 authored by Georg Koppen's avatar Georg Koppen
Browse files

Merge remote-tracking branch 'gitlab/merge-requests/83' into maint-1.1

parents 36d96da7 316d67fe
Pipeline #5026 passed with stage
in 49 minutes
......@@ -187,7 +187,8 @@ unreleased_version_label = "Unreleased"
## Examples:
## - makotemplate("restructuredtext")
##
output_engine = rest_py
#output_engine = rest_py
output_engine = makotemplate(".gitchangelogrst.tpl")
#output_engine = mustache("restructuredtext")
#output_engine = mustache("markdown")
#output_engine = makotemplate("restructuredtext")
......
......@@ -19,15 +19,11 @@ ${section["label"]}
${"~" * len(section["label"])}
% endif
% for commit in section["commits"]:
<%
subject = "%s [%s]" % (commit["subject"], ", ".join(commit["authors"]))
entry = indent('\n'.join(textwrap.wrap(subject)),
first="- ").strip()
<% c = commit["subject"]
if commit["body"]:
c += "\n" + commit["body"].replace("\n\n", "\n")
entry = indent(c, first="- ").strip()
%>${entry}
% if commit["body"]:
${indent(commit["body"])}
% endif
% endfor
% endfor
......
......@@ -77,7 +77,7 @@ python37tormaster:
variables:
BASE_IMAGE: python:3.7
RELEASE: tor-nightly-master-buster
TOR: tor=0.4.6*
TOR: tor=0.4.7*
image: $BASE_IMAGE
script:
- tox -e py37
......@@ -131,3 +131,20 @@ python310:
- tox -e py310
- tox -e integration
allow_failure: true
release_job:
before_script:
- echo "Nothing"
after_script:
- echo "Nothing"
image: registry.gitlab.com/gitlab-org/release-cli:latest
only: [tags]
script:
- echo "Running release job."
release:
name: "Release $CI_COMMIT_TAG"
description: "Created using release-cli"
tag_name: "$CI_COMMIT_TAG"
ref: "$CI_COMMIT_TAG"
milestones:
- "sbws: 1.1.x-final"
Authors
=======
The following people have contributed to Simple Bandwidth Scanner.
Thank you for helping make Tor better.
The following people have contributed to Simple Bandwidth Scanner. Thank
you for helping make Tor better.
* anadahz
* George Kadianakis
* Georg Koppen
* juga
* Matt Traudt
* teor
- anadahz
- George Kadianakis
- Georg Koppen
- juga
- Matt Traudt
- teor
*Last updated: 2020-06-26 on d7a822bf*
......@@ -7,6 +7,613 @@ The format is based on `Keep a
Changelog <http://keepachangelog.com/en/1.0.0/>`__ and this project
adheres to `Semantic Versioning <http://semver.org/spec/v2.0.0.html>`__.
1.2.0
-----
New
~~~
- Docs: Include script on how to release.
- Scripts: Add script to help new releases.
- Add gitchangelog template.
- Add gitchangelog configuration file.
- Docs: Add bwauths list image.
- Relaylist: Keep relays not in last consensus.
Keep the relays that are not in the the last consensus, but are not
"old" yet.
- Util: Add function to know if timestamp is old.
Part of #30727.
Changes
~~~~~~~
- Stem: Set default torrc options.
when connecting to an external tor and they are not already set.
- Generate, cleanup: Use 28 days of measurements.
When generating the Bandwidth File as Torflow, use 28 days of past raw
measurements instead of 5, by default.
Also keep the raw measurements for that long before compressing or
deleting them.
And stop checking whether the compression and delete
periods are valid, without checking defaults first and based on
arbitrary values.
- Stem: Add function to connect or start tor.
Move initialization via existing socket to this new function and start
tor only when it fails.
- Stem, scanner: Change args initializing controller.
to check whether the external control port configuration is set.
There is no need to assert all argument options nor to return the error.
- Config: Add option to connect to external tor.
via control port.
- Circuitbuilder: Remove not used attributes.
and make argument optional.
- Circuitbuilder: Simplify building circuit.
Since sbws is only building 2 hop paths, there is no need to add random
relays to the path, or convert back and forth between fingerprint and
``Relay`` objects.
This will eliminate the circuit errors:
- Tor seems to no longer think XXX is a relay
- Can not build a circuit, no path.
- Can not build a circuit with the current relays.
If a relay is not longer running when attempting to build the circuit,
it will probably fail with one of the other circuit errors: TIMEOUT,
DESTROYED or CHANNEL_CLOSED.
- Scanner: Stop storing recent_measurement_attempt.
because it stores a timestamp for each attempt, which makes state.dat
grow thousand of lines (json).
Closes #40023, #40020
- V3bwfile: Exclude relays without observed bw.
and without consensus bw from scaling.
Part of #33871, closes #33831.
- V3bwfile: Percentage difference with consensus.
- V3bwfile: Calculate hlimit from scaled sum bw.
instead of bw before scaling.
Tests have finally correct value.
For 1 result, only when the cap is 1, the value will be equal to the
rounded bw because the cap does not limit it.
- V3bwfile: Obtain consensus values from last consensus.
- V3bwfile: Round scaled bandwidth after capping.
Make tests pass because the high limit change the expected values,
but the final value still needs to be fixed.
- V3bwfile: Change logic obtaining min bandwidth.
Take either the consenus bandwidth or the descriptor bandwidth if
one of them is missing, do not scale when both are missing and
ignore descriptor average and burst when they are missing.
- V3bwfile: Scale relays missing descriptor bws.
Scale relays without average or observed bandwidth.
Later it will be check what to do if their values are None or 0
- V3bwfile: Stop making mean minimum 1.
- V3bwfile: Calculate filtered bandwidth.
for each relay, calculate the filtered mean for all relays and
calculate the filtered ratio for each relay.
- Scaling: Add filtered bandwidth function.
to calculate the filtered bandwidth for each relay.
- Bwfile: Test KeyValues in a bandwidth file.
Added:
- library to check whether the KeyValues make sense
- test an example bandwidth file
- a command to check an arbitrary bandwidth file
Finally, doing something with all these KeyValues!
(Quarantine day 7th)
- V3bwfile: Count recent relay's monitoring numbers.
using timestamps class.
Also add one more result to the tests data and change the
test accordingly.
- Tests: Remove `_count` from attr.
- Resultdump: Add missing attrs to errors.
- Resultdump: Remove `_count` from attributes.
Tests wont' pass with this commit, they'll be fixed in the next commits
- Relayprioritizer: Count priorities with timestamps.
in RelayPrioritizer:
- Rename recent_priority_list_count to recent_priority_list when
there is no counting
- Rename recent_priority_relay_count to recent_priority_relay
when there is no counting
- Use the timestamps class to manage/count priority lists/relays
- Relaylist: Count measurements with timestamps.
in RelayList:
- Rename recent_measurement_attempt_count to recent_measurement_attempt when
there is no counting
- Use the timestamps class to manage/count measurement attempts
- Relaylist, v3bwfile: Count consensus with timestamps.
in RelayList:
- Rename consensus_timestamps to recent_consensus
- Rename recent_consensus_count to recent_consensus when there is
no counting
- Use the timestamps class to manage/count consensuses
- Remove method not needed anymore
- V3bwfile: Convert datetime to str.
- Resultdump: Use custom json encoder/decoder.
- State: Encode/decode datetimes.
- Json: Create custom JSON encoder/decoder.
to be able to serialize/deserialize datetime in the state file.
- Timestamps: Add module to manage datetime sequences.
- State: Add method to count list values.
Fix
~~~
- Clarify release script dependencies.
- Use rst changelog template.
and put in the same entry commit subject and body removing new lines.
- Correct network stream and filtered bw.
because Torflow is not using them by relay type.
- V3bwfile: network means without relay type.
This reverts commit fc3d3b992ada601a6255f8a6889179abd4b7e55e and partially
reverts a82c26184097bea3ca405ae19773de7c4354a541.
It was a mistake to think torflow was using the means by relay type,
it actually sets the same networks means for all relay types.
Closes #40080.
- Semi-automatic correction of typos.
Closes #33599.
- Tests: Add codespell configuration.
- Tests: Additional security tests.
- CI: Use all tox environments for python 3.8.
- 2nd round of automatic format.
black insists to keep one long line and flake complain, therefore make
flake to ignore it.
- Flake8 errors.
- Reorder imports with isort.
- Reformat all with black.
- Move to declarative setup.cfg.
Also:
- Update versioneer
- And include other source distribution files in MANIFEST.in
- Add project URLs
- Add formatter and linter dependencies and configurations.
- tox: Remove travis, fix python environments
- tox: Remove extra coverage options and add them in .coveragerc.
- Indent by default to 2 except python files.
also uncomment final newline. Can be commented again in case it fails
- V3bwfile: network means by relay type.
Calculate network stream and filtered bandwidth averages per relay
type, to obtain bandwidth weights the same way as Torflow.
Closes #40059.
- Scaling: Return mean if no bw >= mean.
- Scaling: Stop returning 1 as the means minima.
since they are used as the numerator when calculating the ratio and the
rounding already returns a minimum of 1.
- Scaling: Return if there are no measurements.
it should not be the case because the measurements come from
successful results, but otherwise it'd throw an exception.
- Tests: Add bw filtered from results.
- Scaling: round bandwidth filtered.
because Torflow does it.
- Scanner: Return from measure if no helper.
After refactoring in #40041, it was forgotten to return the error in
the case a helper was not found, what can happen in test networks.
Closes #40065.
- Tests: debug log for tests by default.
and fix test that didn't consider that there might be other logs from
other threads.
Closes #33797.
- Scanner: Log times kept.
not only the times that are not kept.
Closes #40060
- CI: Temporal workaround for #40072.
- Relalist: Use the consensus timestamp.
to the relay consensus timestamps list, so that it can be
tested it was in a concrete consensus.
- CI: Exit from integration script.
when any of the commands fail.
- CI: Update Python versions.
Closes #40055.
- CI: Update tor versions.
- System physical requirements.
After fixing #40017, the datadir files are compressed after 29 days and
deleted after 57. However the total used disk space is less than 3G,
leaving 3G as precaution.
Closes #40044.
- Scanner: Return from measure if no helper.
After refactoring in #40041, it was forgotten to return the error in
the case a helper was not found, what can happen in test networks.
Closes #40065.
- Update differences Torflow/sbws.
Closes #40056
- Reorganize Torflow aggregation.
- reorganize sections
- add diagrams and links
- add pseudocode
- remove math
- correct statements
So that it's more accurate and easier to understand.
- Docs: Rename section, add diagrams.
- Separate Torflow/sbws differences.
into a new file.
- Add target to call plantuml.
and generate .svg from .puml files.
Do not add to the html target since the generated svg images are
not deterministic and will change every time `plantuml` is call.
- Separate how scanner and generator work.
in different files and link to each other.
- Add missing new lines.
- CI: Make wget quiet.
to avoid many lines of non useful text the CI.
- Scanner: Rm condition assigning helper.
- Scanner: Move as_entry/as_exit into one function.
since they're similar code
- Scanner: remove relay to measure as helper.
- Scanner: log exit policy when stream fails.
- Relaylist: Remove duplicated can exit methods.
After refactoring and making clear when we were using exit(s) that can
exit to all public IPs (and a port) or only some, refactor them
removing the duplicated code and adding the `strict` argument.
- Add relay measure activity diagram.
- Scanner: extract method on circuit error.
At some point all possible errors should be exceptions.
- Scanner: extract method for not helper case.
- Scanner: extract method to create paths.
because `measure_relay` method is too long, confusing and we have had
several bugs in this part of the code.
- Relaylist: Add methods to obtain exits that.
can exit to some IPs.
To use them in the cases it will be more convenient.
- Relaylist: rename exits_not_bad_allowing_port.
see previous commit
- Relaylist: rename is_exit_not_bad_allowing_port.
see previous commit
- Relaylist: rename can_exit_to_port.
to can_exit_to_port_all_ips, because it's using `strict`, which means
that it allows to exit to all IPs.
It seems more convenient to try first with exits that allow to
exit to some IPs and only try a second time if that fails, because
there are more.
- Resultdump: Check that the error has a circuit.
Because if the error is not a circuit error, it does not have that
attribute.
- Tests: Run integration tests with chutney.
and adapt the tests to pass.
\o/
- Add chutney configuration.
and scripts to run the integration tests with chutney.
It does not replace yet the way integration tests are run.
- Stem: Move torrc option that does not depend on config.
It seems we forgot this option when refactoring in #28738.
- Stem: Remove torrc option that is the default.
to avoid conflict when comparing the options that should be set and the
ones are set, since the SocksPort will be differently in chutney.
- Resultdump: Log if relay was measured as exit.
or entry.
Closes #40048
- Relaylist: Stop measuring relays not in the consenus.
as this might cause many circuit errors.
They're already added to the generator.
Also adapt the number in test_init_relays.
- Sphinx warnings when creating documentation.
This should give us at least a clean html, text, and man build
experience.
Closes #40036.
- Add forgotten image from consensus health.
It was referenced by 6e6a8f3ba534cbd93b830fe3ffd5ce40abe8e77d. Since that
image was wrong, created a new screenshot from the current "past 90
days" at consensus-health.tpo.
- Stem: Add possible exception cause.
- Stem: Remove unused code.
- Stem: Exit on failure connecting to control port.
because when trying to connect to an external tor (chutney), it does
not make sense to start own tor.
Also log how the connection has been made.
- Update values in config_tor.rst + clean-up.
Closes #40035.
- Update default values in man_sbws.ini.rst.
Closes #40034.
- Clean up config.rst.
Closes #40033.
- Scanner: Retry to measure exit as exit.
if it fails to be measured as entry.
Mayb closes: #40029.
- Relaylist: Comment on IPv6 exit policy.
that could be also checked, increasing the chances that the exit can
exit to our Web servers.
But if it could not, then we need to retry to measure it as 1st hop.
- Config: Increment circuit build timeout.
setting it to the default, 60secs.
Since many relays fail to be measured cause of circuit timeout.
Maybe closes #40029.
- Bump bandwidth file version to 1.5.0.
after removing KeyValue recent_measurement_attempt_count in #40023.
Changed also torspec, issue #20.
- V3bwfile: Tor version added in bandwidth v1.4.0.
since, by mistake, the bandwidth file version here was never updated
to v1.5.0.
This patch only changes the constants names, but logic remains the same.
Related to torspec#35.
- Add the bwauths timeline wiki.
Closes #40013.
- Add bwauthealth tool.
- Add consensus health page.
about bwauths measured relays.
- Move consensus weight to top.
and explain what to check.
- V3bwfile: Take all measurements when IP changes.
Previously, when a relay changes IP, only the measurements with the
last IP were considered.
Relays with dynamic IP could get unmeasured that way.
Now, all the measurements are considered.
- V3bwfile: Avoid statistics without data.
If mean or median argument is empty, they throw an exception.
This can happen when the scanner has stopped and the result is
stored as successful without any downloads.
- No need to use Travis anymore.
- Clarify branch to use when contributing.
- Maint: Fix linter error after merging #29294.
- Tests: Stop converting boolean key to int.
Conversion only happens when parsing a bandwidth file in the
integration tests.
- Relaylist: filter out private networks.
when checking exit policies to know whether an exit can exit to a port.
- Update authors.
- Replace docs links from Github to Gitlab.
- Update reviewers.
- Replace Github review process to Gitlab.
Replace also Github terminology to Gitlab.
- Replace Trac, ticket by Gitlab, issue.
- Replace links from Trac to Gitlab.
- Start using release script later.
Change the version from which the release script is used.
Also explain the prefixes used in the commits.
Closes #29294
- Scripts: Clarify the scope of the script.
it should not take more effort than solving self-sbws issues.
- Scripts: Reformat sentence.
- Scripts: Stop bumping to next prerelease version.
since it is now managed automatically by versioneer.
Instead, suggest creating a "next" maintenance branch.
But stop using `-` and `.` characters in it, to type it faster, since
most of the new branches will be based on it.
- Scripts: Stop releasing from -dev0 version.
since now sbws version is calculated from last release tag.
- Scripts: Stop changing version in __init__
Since it is now done by `versioneer`.
- Scripts: Change Github by Gitlab.
releases can live now in gitlab.tpo, instead of github.com and
there is no need to check them since Gitlab is FLOSS and gitlab.tpo is
hosted by Tor Project.
Also, stop assuming which is the current branch and remote and do not
push. Instead guide the maintainer to do it.
- CI: Add .gitlab-ci.yml to run tests in Gitlab.
- Relaylist: Check exit to all domains/ips.
When an exit policy allows to exit only to some subnet, it is not
enough to check that it can exit to a port, since it can, but it might
not be able to exit to the domain/ip of the sbws Web servers.
To ensure that without having to check whether it can exit to a
specific domain/ip, we can query the exit policy with `strict`.
Closes #40006. Bugfix v1.0.3.
- V3bwfile: Count relay priority lists.
and measurement attempts from all the results.
Until they get properly updated.
Also change dates in tests, so that timestamps are counted correctly
- Recomment maint-1.1 for production.
- Recommend using a CDN,
add link to it and rephrase some sentences.
- Increase RAM required.
ahem, because of all json it has to manage in memory.
- Recommend pip only for development.
or testing and add links.
- Update supported Python versions.
- Comment on Debian/Ubuntu releases.
because sometimes the package might not be in Debian stable or testing
and we are not checking Ubuntu releases.
- Tests: Remove all the `\t` in torrc files.
at the beginning of the line and in empty lines. They are not needed.
- Tests: Create new authority keys.
because they expired.
They will expire again in a year.
Implementing #33150 and using chutney would avoid to update keys.
Closes #34394.
- V3bwfile: linter error with new flake version.
- Add differences between Torflow and sbws.
Closes #33871.
- Update/clarify Torflow aggregation.
- Docs: Remove unneeded linter exception.
- Docs: Move torflow scaling docstring to docs.
so that it has its own page as it is too long as docstring and is
harder to write latex with the docstring syntax.
- Unrelated linter error.
- V3bwfile: Remove unneeded minimum 1.
since rounding already returns 1 as minimum.
- V3bwfile: Use cap argument to clip scaled bw.
Make test pass, though the value is not correct since it needs to be
rounded after clipping
- V3bwfile: cap is never None.
- V3bwfile: Warn about None bandwidth values.
since they are probably due a bug.
- Check that log prints a number.
and not a list of timestamps.
- Assert that caplog messages were found.
- Explain changes in the previous commits.
- Tests: Check the files generated in test net.
Test that the results, state and bandwidth file generated by running
the scanner and the generator in the test network are correct.
- Tests: Add tests loading results.
in ResultDump and incrementing relay's monitoring KeyValues.
- Tests: Add results incrementing relays'
monitoring KeyValues.
- V3bwfile: Stop calculating failures with 0 attempts.
- Relaylist: Count recent relay's monitoring numbers.
using timestamps class.
Additionally:
- fix: relayprioritizer: Replace call relay priority
- fix: scanner: Replace call relay measurement attempt
- State: Let json manage data types.
Since state uses json and json will raise an error when it can't
decode/encode some datatype.
- State: Read file before setting key.
Otherwise, if other instance of state set a key, it's lost by the
current instance.
Bugfix v0.7.0.
- Tests: Test state file consistency.
Test that two different instances of state don't overwrite each other.
This test don't past in this commit, will pass in the next bugfix.
Bugfix v0.7.0, which claimed 100% test coverage on state.
- Tests: linter error cause missing nl.
- Relaylist: Update relay status before consensus.
Update relay status before updating the consensus timestamps
Timestamps that are not old yet were getting removed because the
document.valid_after timestamp was still the one from the previous
consensus.
Closes #33570.
- Tests: Test the number of consensus in Relay.
This test does not pass in this commit, but in the next bugfix.
- Relaylist: Use is_old fn removing consensus.
since the logic is the same and the there were two bugfixes on the
same logic.
- Relaylist: Use seconds removing consensuses.
by default days is passed to timedelta, what was making the oldest
date be thousands of days in the past.
Bugfix 1.1.0.
- Tests: Add relaylist test.
Tests don't pass in this commit, they're fixed in the next commits.
- Tests: Add mocked controller fixture.
to be able to unit test all the code that needs a controller.
- Tests: Add test for remove old consensus ts.
Tests don't pass in this commit, it's fixed in the next commits.
- Timestamp: measurements period is in seconds.
by default days is passed to timedelta, what was making the oldest
date be thousands of days in the past.
- Timestamp: Old timestamps are minor than older.
Old timestamps are minor than the older date, not major.
- Relaylist: Stop passing argument to self.is_old.
- Tests: Add test timestamp.is_old.
The tests don't pass in this commit, it's fixed in the next ones.
- V3bwfile: Reformat to don't get flake8 errors.
Part of #30196
- V3bwfile: Move keys to correct constant.
Part of #30196.
- V3bwfile: Add comment about bwlines v1.3.
Part of #30196.
- V3bwfile: Add tor_version KeyValue.
- Create new KeyValues constants for the new v1.5.0 KeyValues
- Instantiate State in Header.from_results so that there is no need
to create new methods for all the header KeyValues that are read
from the state file
- Add tor_version to the kwargs to initialize the Header
- Write tor_version in the state file when the scanner is started
- V3bwfile: Add constant for ordered key/values.
to build the list of all keys from it and ensure no key is missing.
- V3bwfile: Reformat to don't get flake8 errors.
After the automatic constants renaming, fix the flake8 errors by
reformatting automatically with `black`, only the lines that had
errors.
Part of #30196
- Document why ersioneer to obtain version.
- Add at build time the git revion to version.
Instead of having a hardcoded version, calculate the version at build
time making use of `git describe --tags --dirty --always`.
This way, even if the program is not running from inside a git
repository it still can know which was the git revision from the
source it was installed from.
If the program is launched from a path that is a git repository, it
does not gives the git revision of that other repository.
If's also able to get the version when installed from a tarball.
It does not add the git revision when it's being install from a git
tag.
`versioneer` external program is only needed the first time, because
it copies itself into the repository. So it does not add an external
dependency.
There're no changes needed to the `--version` cli argument nor to the
code that generates the bandwidth file, since they both use the
variable `__version__`.
The version previous to this commit was `1.1.1-dev0`, after
this commit, it becomes `1.1.0+xx.gyyyyyyyy`, ie. xx commits after
`1.1.0` plus the git short hash (yyyyyyyy).
- Tests: Test maximum retry delta in destination.
- Destination: Replace constant name.
to make it consistent with others and shorter.
Part of #33033.
- Destination: Set maximum to retry a destination.
otherwise the time to retry a destination could increase too much,
as it's being multiplied by 2, and a destination would not recover.
Patch submitted by tom.
- Relaylist: linter error after after merge.
Fix linter error after merging #30733 and #30727.
- CI: Cache pip, run tox stats after success.
and do not require sudo.
- CI: Test all supported python versions.
As in chutney and stem:
- Test all supported python versions
- Test all supported tor versions
Differences between chutney, stem and sbws:
- in sbws we run directly, not an script that calls tox
- we're not using chutney for integration tests (yet) and therefore we're not testing it with different networks
- we don't have shellcheck tests
- we don't support osx nor windows
- Relaylist: Update the relays' descriptors.
when fetching new consensuses.
Part of #30733.
- Globals: Fetch descriptors early.
and useless descriptors, so that sbws detect early changes in the relay
descriptors and continue downloading them even when Tor is idle.
Other
~~~~~
- Wip: rm me, temporally change release url.
to personal fork, to test the release process
- Fixup! minor: Change info logs to debug or warning.
- Major: Change default log level to info.
also change formatting to show thread.
- Minor: Change log warning to info or debug.
when it contains sensitive information.
- Minor: Change info logs to debug or warning.
when they contain sensitive information, eg. Web server or are too
verbose for the debug level.
Also add log to indicate when the main loop is actually started.
- Revert "fix: stem: Remove torrc option that is the default"
This reverts commit 15da07d6a447d8310354124f6020b4cf74b75488.
Because it's not the default. No additional changes are needed in the
tests.
Closes #40064.
- Minor: scanner: Change logic creating the path.
When the relay is not an exit, instead of choosing exits that can
exit to all IPs, try with exits that can exit to some IPs, since the
relay will be measured again with a different exit in other loop.
When the relay is an exit, instead of ensuring it can exit all IPs, try