Conversation
|
Looks like tests are failing due to a networking issue connecting to MITRE's artifact repository. Will have to investigate. |
|
Ah, it seems MITRE's changes to NPM are breaking the build here, because |
alilleybrinker
force-pushed
the
alilleybrinker/minor-fixups
branch
2 times, most recently
from
b6f9b51 to
327469b
Compare
|
I figured out the right incantation to fully clear Artifactory mentions, so packages should now all pull from NPM properly. |
|
Huh, for some reason the tests were hanging after an hour and a half with a MongoDB error that was not causing the test infrastructure to give up. That tells me:
|
|
MongoDB issue observed in prior runs appears to be transient and non-reproducible on my local device. I'm hopeful it's just a GitHub Actions issue that will disappear on a retry now that it's been about a week. |
|
Nope, errors still appearing. Mongo is claiming that it's trying to use a session that's ended. Time to dive into debugging then! |
|
Possibly resolved the issue, which appears to have been due to a race condition due to missing |
|
Checks cancelled due to GitHub Actions outage. Will rerun when the outage ends. |
|
Ah, I'd also not actually pushed the fix commit to the branch. That's now done. We'll see if GitHub Actions actually runs. |
|
Progress! There is now a new baffling error message, this time from the internals of our MongoDB library where somehow a variable called |
Put it into the .github folder instead of the top-level folder of the repo. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
This makes some basic Python code quality improvements to the 5.0 upload script. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
This corrects some broken Markdown in the test README file, and also removes some incorrect information about pre-push hooks that don't exist. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
The `action:test` script previously did not set the `NODE_ENV` environment variable, which caused warnings to be emitted by winston, our logging library, indicating that it had nowhere to send the logs it was producing. This is because the log initialization code in cve-services checks `NODE_ENV` to figure out how to configure the "transport" (Winston's term for a log sink), so without `NODE_ENV` being set, the transport is never configured. This fix just sets the environment variable, causing the warnings to disappear. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
Did three things: - Ran `npm run clean` to identify unused deps. Confirmed each is unused manually before removing. - For two false positives, modified the `clean` command to ignore them. - Renamed the `clean` command to `depcheck`, to more accurately describe what it's doing. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
Several NPM scripts were duplicative of others, and have been removed. This also reorders scripts to make sure the ordering is more consistent. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
This info said to run a command that doesn't exist. This also introduces a guide for running integration tests. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
Felt odd to have a script sitting in the root of the repo. This relocates it to `src/scripts/`, which is technically odd because it's a Python script not a JavaScript one, but this is already a grab-bag of miscellaneous scripts that get run during development, so I think it's fine. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
This also cleans up the Markdown formatting of the top-level project README.md. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
Result of running `npm update` to get the latest SemVer-compatible versions of existing dependencies. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
This fixes a race condition present in the existing populate script caused by missing `await`. For consistency, it transitions the entire file to an ECMAScript module, which permits top-level await, and then updates all uses of Mongoose APIs to 1) make sure we're always `await`-ing any Promises, and 2) only use `await` rather than any alternative callback-oriented APIs. This appears to have resolved the race locally, which previously permitted early-closure of the connection to the MongoDB database. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
Scripts appear under `src/`, but aren't part of the application itself and don't need to be coverage-tested. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
alilleybrinker
force-pushed
the
alilleybrinker/minor-fixups
branch
from
9248673 to
b05f550
Compare
alilleybrinker
requested review from
david-rocca
and removed request for
david-rocca
This file is a common standard used by many Node version-management tools that helps ensure developers are using the "current" version of Node recommended for the project. I've matched the version here to the one set in the Dockerfile, so it'll match the tests run in CI. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
This updates to the latest compatible dependency versions for all existing dependencies. Signed-off-by: Andrew Lilley Brinker <abrinker@mitre.org>
|
@david-rocca CI is failing, but this PR doesn't actually modify any application code, so these are either transient failures or failures on |
1 participant
Did these concurrent with my work on the new authz API. The fixups are organized by-commit, so reviewing by-commit might be easier. Opted to bundle them into a single PR to make review simpler; happy to split them out if that's preferable.
The changes are:
.github/folder.NODE_ENV=testconsistently on unit tests.src/scripts/replSetfor MongoDB to work correctly.