| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
| |
I just had a situation where the backtrace might have been helpful,
arti#2538, https://gitlab.torproject.org/Diziet/arti/-/jobs/1539809.
I was surprised to find we don't seem to turn this on and never have!
|
| |
|
|
|
| |
This test is known to fail, and we have had a ticket for it for a week
now. It should be re-enabled if and when it works again.
|
| | |
|
| | |
|
| |
|
|
|
|
| |
Mostly to get logging improvements chutney!124.
Also work around a potential race condition, though I don't think we've
observed it in CI chutney!125
|
| |
|
|
|
| |
Primarily to get recent improvements to bootstrap checking and
reporting.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
Including these would introduce flakiness, until arti#2400 is resolved.
This also adds a general mechanism for omitting specific features in the
arti-extra build.
|
| |
|
|
|
|
|
|
|
|
|
| |
Previously we enabled features piecemeal to avoid enabling `rpc`, which
resulted in breakage when running under shadow. Experimentally, this
appears to no longer be the case.
Enabling everything in `full` and `experimental` gives us at least a
little bit of testing of those features (that they don't break the build
or other usage), even if we don't specifically exercise them (yet) in
the CI tests.
|
| |
|
|
|
| |
Test the arti-extra bin, which includes most of `full` and
`experimental` features, too.
|
| |
|
|
|
| |
mypy needs it to correctly type-check the chutney test scripts, which
now import chutney as a module
|
| |
|
|
|
|
|
| |
chutney replaced Node.socksport (a single optional port) with
Node.socksport_endpoints (a sequence of bound addresses and ports).
Update for compatibility, and test all of the exposed endpoints where
relevant.
|
| | |
|
| |
|
|
|
|
| |
See: arti#2367
See: arti!3703
See: arti!3690
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch changes the behaviour of our `cargo-audit` CI job:
- All non-merge requests (pushes, manual jobs, scheduled jobs, tags, etc.) will
continue to run with allow_failure set to false.
- When a push is made, where there are changes to `**/Cargo.lock`, we
continue to NOT allow failures.
- When a push is made, where there are NO changes to `**/Cargo.lock`, we
now allow failures.
Specify workflows for Arti in our GitLab CI configuration:
- We currently run our jobs on pushes, and upon tags.
This was discussed in the Network Team meeting on 2026-02-10.
|
| |
|
|
|
|
|
| |
As agreed at our last team meeting.
See
https://gitlab.torproject.org/tpo/core/arti/#minimum-supported-rust-version
|
| |
|
|
|
|
| |
All (publishable) crates should explicitly configure what features to (not) document in docs.rs.
This to avoid our documentation containing references to public API specifications which yield "given resource not found"
due to them being behind a feature flag not being published to docs.rs by default.
|
| | |
|
| |
|
|
| |
Bump chutney, and migrate to CLI changes introduced in chutney!101
|
| | |
|
| |
|
|
|
|
|
| |
Change the script name to use hyphens, not underscores.
Make the script name and the CI job name the same, except the script
has -arti at the end in case we add more similar checks.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
Swap the symlink direction: now we have list_crates.py (the python
module) with a symlink list-crates pointing to it. This allows the
module to be imported but the command to be kebab case.
The checking tooling doesn't understand this special arrangement.
Lots of references to update!
|
| | |
|
| |
|
|
| |
We had both maint/docker-... and maint/docker_...
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Use kebab case, and also add the word deps. Otherwise it might refer
to files or something.
|
| | |
|
| |
|
|
|
|
| |
The job was called `check-targets` and the script was called
`cargo_check_target`. Make them match. Use kebab case, and the
plural (it checks multiple targets).
|
| |
|
|
| |
As per the docs for that script.
|
| | |
|
| |
|
|
|
| |
it doesn't depend on anything else in the pipeline, so avoid skipping it
when other jobs fail
|
| |\
| |
| |
| |
| |
| |
| | |
build website for and deploy to static component
Closes tpo/tpa/team#42288
See merge request tpo/core/arti!3547
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
this uses the static-shim-deploy template to deploy the arti website to
arti.torproject.org, and tweaks the docusaurus baseUrl accordingly
it requires a STATIC_GITLAB_SHIM_SSH_PRIVATE_KEY CI variable which is
set up in project settings and contains the private key which allows
deployment of assets to static component web hosts
it also deploys a "_redirects" file to gitlab pages to ensure that the
old URLs are redirected
as previously, the website build and deployment jobs are only run in
(nightly) scheduled pipelines
|
| |/
|
|
|
|
|
|
|
|
|
|
| |
Because Rust 1.91.0 makes it economically unfeasiable to run `cargo doc`
oneselves, it might wise to keep it as an artifact in `rust-dev-recent`
and `rust-nightly`, which generate the docs anyways, so we can keep us
away from having to run the expensive command when reviewing (larger)
changes to documentation.
For now, these artifacts will expire within a week.
Maybe this should get discussed, because on my local machine, they are
over 2GB large?
|
| |\
| |
| |
| |
| | |
ci-check-msrv: Add check for MSRV
See merge request tpo/core/arti!3405
|
| | |
| |
| |
| | |
Checks if each crate has rust-version property in Cargo.toml.
|
| | | |
|
| | |
| |
| |
| |
| | |
This downgrades the container image for the 'cargo-miri' job, and
updates some adjacent comments.
|
| |/ |
|
| |
|
|
|
|
|
|
|
|
| |
This also updates some needs statements that existed to order jobs that
did not actually have a dependency on each other.
The idea here is to run everything with as much parallelism as possible,
and if that parallelism causes problems, we should ideally solve it by
adding more runner capacity, or taking a closer look at what the actual
problem is.
|
| |
|
|
|
| |
I am interested in enabling opentelemetry in chutney tests to enable
easier debugging of failures that pop up in CI.
|