Skip to content

TEMPORARY: testing solution_analytics Slack notifications - #54

Draft
fivetran-catfritz wants to merge 21 commits into
mainfrom
test/slack-notifications
Draft

fivetran-catfritz wants to merge 21 commits into
mainfrom
test/slack-notifications

Conversation

@fivetran-catfritz

Copy link
Copy Markdown
Contributor

Throwaway PR to validate the branch-head PR lookup tier in solution_analytics#slack-pipeline-notifications's resolve_pr_header.sh. Not meant to be merged - close once the Slack notification tests are done.

fivetran-catfritz and others added 21 commits October 5, 2026 15:14
…e-notifications)

Adds a throwaway .buildkite/pipeline.yml to bypass the main-branch bootstrap
in fivetran/infrastructure and pull execute_buildkite_pipeline.sh from the
branch under test, plus test_branch in test_scenarios.yml so the generated
per-warehouse steps also pull run_warehouse_tests.sh/notify_slack*.sh from it.
Revert both once validated.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Hand-writes the postgres + notify steps directly instead of calling
execute_buildkite_pipeline.sh, since it has no "just one warehouse" option
and this is throwaway. Drops the now-unused test_branch setting since this
path no longer goes through the generator that reads it.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Local testing of resolve_slack_mention.sh passed with real network/gcloud
calls, but the live build result showed a plain "@fivetran-catfritz" instead
of a real Slack mention - meaning the yq lookup fell through on the actual
agent despite the mapping file being confirmed present there. Isolating yq's
version and raw query output directly, unsuppressed, to see why.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…group fallback

Lets us validate live whether Slack's User-type Workflow variable accepts a
usergroup ID, without needing a real unmapped-author commit that also carries
this throwaway test pipeline.yml.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
solution_analytics PR #16 merged - final smoke test of the real merged
scripts, with a normal resolvable commit instead of the forced-bogus one
used to validate the author fallback.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…n all-pass build

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…e check

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Instead of a separate debug step, the snowflake step now detects a late
manual retry inline (BUILDKITE_RETRY_COUNT > 0 and the real notify step's
state already "finished") and re-runs notify_main_ci_result.sh directly as
part of its own command. Also simulates the retry actually passing, to
validate the "it was just a flake" scenario end to end.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…rmit_on_passed PR

Drops the inline retry-detection experiment (broken by BuildKite's
upload-time variable interpolation on commands: blocks - parked for later).
Points the test harness at slack-notify-permit-retry (PR #18) and adds
permit_on_passed: true to this hand-written test pipeline's own notify step,
to validate the fix directly: manually retry the already-passed notify step
and confirm it's allowed and re-sends a message.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

This branch has not been deployed

No deployments
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