Skip to content

ci(e2e-prod): full live suite on cron+on-demand, not every UI build (stop prod self-DoS)#208

Merged
mastermanas805 merged 1 commit into
mainfrom
ci/right-size-e2e-prod-cadence
Jun 7, 2026
Merged

ci(e2e-prod): full live suite on cron+on-demand, not every UI build (stop prod self-DoS)#208
mastermanas805 merged 1 commit into
mainfrom
ci/right-size-e2e-prod-cadence

Conversation

@mastermanas805
Copy link
Copy Markdown
Member

Running the full live-provision suite on every push:[main] + every api deploy fired it dozens of times on a busy day → accumulated real customer DBs on shared prod infra faster than reaps cleaned them → prod provisioning degraded (the vector live test went 10s → 2-min timeout within one session).

Remove push:[main]; keep cron (*/30) + workflow_dispatch + an on-demand repository_dispatch. Cheap per-build coverage is unchanged: e2e-pr-smoke (contract-only) on every web PR + the api webhook-injection unit test on every api PR. Paired with reverting the api per-deploy dispatch.

🤖 Generated with Claude Code

…build)

Running the full live-provision suite on every merge to main (push:[main]) +
every api deploy fired it dozens of times on a busy day, accumulating real
customer DBs on shared prod infra faster than reaps cleaned them — degrading prod
provisioning (the vector live test went 10s→2min timeout over one session).
Remove push:[main]; keep cron(*/30) + workflow_dispatch + an on-demand
repository_dispatch. Cheap per-build coverage stays: e2e-pr-smoke (contract-only)
on web PRs + the api webhook-injection unit test on api PRs.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@github-actions
Copy link
Copy Markdown

github-actions Bot commented Jun 7, 2026

size-limit report 📦

Path Size
dist/assets/index-CgN1XV4b.js 161.99 KB (0%)
dist/assets/index-BsJUZYRr.css 6.13 KB (0%)

@mastermanas805 mastermanas805 merged commit ee1be4c into main Jun 7, 2026
18 checks passed
@mastermanas805 mastermanas805 deleted the ci/right-size-e2e-prod-cadence branch June 7, 2026 11:55
mastermanas805 added a commit that referenced this pull request Jun 7, 2026
…s (de-flake) (#209)

The live-provision (vector/cache/nosql/db) + claim-deploy legs do real prod
provisioning + an assert-usable connect from the CI runner. Normally ~14s, but
under prod contention they intermittently spike past the 120s default and time
out — and since both files are serial groups, one slow flow aborts the rest
(seen as vector/cache 2min timeouts that pass on retry). Raise the per-test
budget to 180s so transient prod latency stops redding the suite; the test still
verifies the flow WORKS, and retries cover the rest. Pairs with the cadence
right-size (#208/#277) that removes the self-inflicted load.

Co-authored-by: Claude <noreply@anthropic.com>
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.

2 participants