fix(telemetry): use the stable pgflow collector endpoint - #692
Merged
Merged
Conversation
The 0.17.1 CLI and database senders use a workers.dev hostname without the account subdomain, so telemetry cannot reach the collector. Point both senders at telemetry.pgflow.dev while preserving failure handling and privacy safeguards. Add a generated function-only upgrade migration without rewriting released history or enabling telemetry for opted-out databases. Update regression checks, upgrade guidance, and patch changesets. Collector deployment and release remain separate prerequisites tracked in #691.
🦋 Changeset detectedLatest commit: a99d146 The changes in this PR will be included in the next version bump. This PR includes changesets to release 5 packages
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
|
View your CI Pipeline Execution ↗ for commit a99d146
💡 Verify your cache is correct by running tasks in a sandbox. Read docs ↗ ☁️ Nx Cloud last updated this comment at |
jumski
commented
Oct 2, 2026
jumski
left a comment
jumski
left a comment
Contributor
Author
There was a problem hiding this comment.
See inline comments.
Comment on lines
+118
to
+127
| ## Upgrading from 0.17.1 | ||
|
|
||
| pgflow 0.17.1 pointed both senders at `https://pgflow-telemetry.workers.dev`, a hostname that does not resolve. The corrected endpoint is `https://telemetry.pgflow.dev`. | ||
|
|
||
| - Databases get the corrected daily sender through a new pgflow migration that replaces `pgflow_telemetry.report()`. Update pgflow, rerun `pgflow install` to copy the migration, and apply it with your normal Supabase migration workflow. Databases that disabled telemetry stay disabled: the migration replaces only the function and never reschedules the reporting job. | ||
| - The corrected installation sender ships in the pgflow CLI patch release; update the CLI you invoke (`npx pgflow@latest`). | ||
| - Deploying the collector endpoint alone does not repair 0.17.1: its senders keep using the unresolvable hostname until you upgrade. | ||
|
|
||
| Until the upgrade lands, daily reports from 0.17.1 databases fail silently, exactly like any other send failure: no error surfaces in your database, and the day is not retried. | ||
|
|
Contributor
Author
There was a problem hiding this comment.
this is not needed at all
Review feedback: the sentence restated existing failure behavior and added no upgrade guidance. The upgrade bullets already cover the migration, CLI update, and endpoint prerequisite.
Contributor
🔍 Preview Deployment: Website✅ Deployment successful! 🔗 Preview URL: https://pr-692.pgflow.pages.dev 📝 Details:
_Last updated: _ |
Contributor
🚀 Production Deployment: Website✅ Successfully deployed to production! 🔗 Production URL: https://pgflow.dev 📝 Details:
Deployed at: 2026-10-02T08:13:21+02:00 |
This was referenced Oct 2, 2026
jumski
added a commit
that referenced
this pull request
Oct 2, 2026
#695) ## Summary - Redirect browser GET requests to `https://pgflow.dev/reference/telemetry/` with HTTP 302; leave POST telemetry ingestion unchanged. - Keep other HTTP methods at 405 and check that redirects/rejections do not write Analytics Engine points. - Declare `telemetry.pgflow.dev` as the collector's Worker Custom Domain so Wrangler manages DNS and TLS at deployment. ## Checks - `pnpm nx test-env:fresh edge-worker` passed. - `pnpm nx run @pgflow/telemetry-worker:test` passed: 32 tests. - Nx reports only `@pgflow/telemetry-worker` affected. - `pnpm nx affected -t test typecheck lint --base=HEAD` passed before commit. - `pnpm --dir apps/telemetry-worker exec wrangler deploy --dry-run` passed with the existing `PGFLOW_TELEMETRY` Analytics Engine binding. - `git diff --check` passed; the committed tree matches the checked files. ## Deployment This PR does not deploy Cloudflare infrastructure. After merge, run the collector deployment from main with a token authorized for the Worker and Custom Domain. Keep request logging disabled. GET redirects to the docs; a valid JSON POST returns 204. A production smoke-test POST writes a real Analytics Engine point. No npm/JSR release or changeset is needed: the collector is private and 0.17.2 senders already use this hostname. Related: #691/#692 corrected the sender endpoint; this PR declares the hosting route and browser behavior.
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
https://telemetry.pgflow.dev, replacing the non-resolving hostname shipped in 0.17.1.@pgflow/coreandpgflowwithin the existing fixed version group.Closes #691
Deployment and release gates remain open
This PR delivers the code portion of #691, not its full operational acceptance. No Cloudflare/DNS changes, collector deployment, production ingestion, merge, package publication, or registry verification occurred in this delivery. Complete and record those issue requirements separately; the issue-closing link is not evidence of completion.
Make the custom-domain endpoint live before publishing corrected senders. Existing 0.17.1 databases need the new migration, and CLI installations need the corrected version. No audit replay/backfill or recovery of lost telemetry is included.
Checks
9b106a9f6: 15 checks pass, one skips, none fail. Windows package smoke passed on the user-authorized rerun without code changes;gh pr checks --watchcompleted successfully.Resolved CI infrastructure failure
Windows package smoke failed while downloading the Supabase CLI Windows archive, before any package smoke test ran:
The user explicitly authorized a rerun of the failed job. Attempt 2 passed in 1m36s at the unchanged head. No code repair was needed; the CI gate is clear.
Local evidence qualifications
The broad quality run initially failed only at
demo:testwith a host-provided Groq key: the upstream API reported model unavailability or missing access. The focused reproduction returned the same error; the keyless CI-equivalent run passes. This PR does not change that dependency or establish that the model is retired. Pre-push hooks also ran in the keyless environment.The initial Changesets status check could not see the untracked changeset; the staged review and normal pre-push Changesets checks now pass. The upgrade check log records successful enabled/opted-out checks, but does not retain its runnable script.