Skip to content

ci: run dependency review and workflow lint on pull requests - #174

Merged
pjhartlieb merged 2 commits into
devfrom
ci/dev-security
Nov 9, 2026
Merged

pjhartlieb merged 2 commits into
devfrom
ci/dev-security

Conversation

@pjhartlieb

Copy link
Copy Markdown
Contributor

Adds dev-security.yml, which calls release-workflows dev-security.yml at v1.1.0 (pinned by commit SHA). On every pull request it runs dependency review and the workflow lint (actionlint, zizmor), the security checks publishing policy rule 3 requires on dev. Once this merges, the dev ruleset requires dev-security / dependency-review and dev-security / workflow-lint.

It also fixes what zizmor reports in the existing workflows, so the new check starts green:

  • cla.yml: both triggers marked as reviewed. cla-reusable.yml never checks out pull request code.
  • publish.yml: the same-repository workflow calls marked as reviewed.
  • benchmark.yml (where present): checkouts no longer persist credentials.
  • dependabot.yml: version updates wait 7 days after a release is published (cooldown), so a compromised release has time to be caught first. Security updates are not delayed.

Locally: actionlint 1.7.12 clean, zizmor 1.30.1 (the version the check runs) no findings across the repository.

Calls release-workflows dev-security.yml (v1.1.0), the security checks policy rule 3 requires on dev. Fixes what zizmor reports: the CLA triggers are marked as reviewed (cla-reusable.yml never checks out pull request code), benchmark checkouts drop persisted credentials, and publish.yml's local workflow calls are marked as reviewed (same-repository calls on stable).
zizmor dependabot-cooldown (the workflow-lint check): version updates now wait 7 days after a release is published, so a compromised release has time to be caught and pulled before it is proposed here. Security updates are not delayed.
@github-actions

github-actions Bot commented Nov 9, 2026 •

Copy link
Copy Markdown

DNS Resolver Benchmark

this PR @ a4a45b0 against dev @ ab2568b, base → head, median of 3 interleaved rounds. Bold is significant across all 56 comparisons (Benjamini-Hochberg at 1%); the rest is noise.

regime qps unanswered p99 attempts/query
ceiling/1-resolver 20,546 → 20,128 (-2.0%) 0.00% → 0.00% 0.971s → 0.991s (+2.1%) 1.00 → 1.00
pool/32-resolvers 20,337 → 20,338 (+0.0%) 0.00% → 0.00% 0.981s → 0.980s (-0.0%) 1.00 → 1.00
pool/32-persistent 21,787 → 21,892 (+0.5%) 0.00% → 0.00% 0.915s → 0.911s (-0.5%) 1.00 → 1.00
impaired/10%-loss 219 → 219 (-0.0%) 0.00% → 0.00% 7.313s → 7.314s (+0.0%) 1.11 → 1.11
impaired/20%-REFUSED 2,669 → 2,697 (+1.0%) 0.00% → 0.00% 0.749s → 0.741s (-1.0%) 1.24 → 1.24
impaired/50ms-latency 618 → 617 (-0.2%) 0.00% → 0.00% 3.234s → 3.239s (+0.2%) 1.00 → 1.00
impaired/20%-then-2.5%-dead-names 263 → 233 (-12%) 5.43% → 5.43% 10.223s → 11.125s (+8.8%) 1.05 → 1.05
All metrics
regime p50 p100 sockets conntrack
ceiling/1-resolver 0.529s → 0.535s (+1.2%) 0.973s → 0.993s (+2.1%) 42 → 42 (+0.0%) 14,614 → 12,688 (-13%)
pool/32-resolvers 0.529s → 0.530s (+0.2%) 0.983s → 0.983s (-0.0%) 73 → 74 (+1.4%) 19,217 → 19,399 (+0.9%)
pool/32-persistent 0.491s → 0.491s (-0.0%) 0.918s → 0.913s (-0.5%) 42 → 42 (+0.0%) 29 → 30 (+3.4%)
impaired/10%-loss 3.316s → 3.315s (-0.0%) 9.128s → 9.128s (+0.0%) 46 → 46 (+0.0%) 2,105 → 2,090 (-0.7%)
impaired/20%-REFUSED 0.419s → 0.417s (-0.4%) 0.749s → 0.741s (-1.0%) 46 → 46 (+0.0%) 2,203 → 2,173 (-1.4%)
impaired/50ms-latency 1.845s → 1.848s (+0.2%) 3.234s → 3.239s (+0.2%) 46 → 46 (+0.0%) 865 → 636 (-26%)
impaired/20%-then-2.5%-dead-names 3.762s → 3.731s (-0.8%) 11.396s → 12.878s (+13%) 39 → 39 (+0.0%) 2,971 → 2,950 (-0.7%)

@pjhartlieb pjhartlieb closed this Nov 9, 2026
@pjhartlieb pjhartlieb reopened this Nov 9, 2026
@pjhartlieb
pjhartlieb merged commit 9554347 into dev Nov 9, 2026
37 of 38 checks passed
@pjhartlieb
pjhartlieb deleted the ci/dev-security branch November 9, 2026 17:29
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant