E2E Chain Validation has failed nightly since 2026-09-02: pinned bulletin-deploy 0.15.0 predates the DotNS redeploy fix

Hello from the bulletin-deploy side — flagging this because you likely have not seen it, and the fix is one line.

What is happening

E2E Chain Validation has failed every night since 2026-09-02:

2026-09-01 03:31  success   ← last green
2026-09-02 03:29  failure
2026-09-03 04:41  failure
2026-09-04 03:28  failure
2026-09-05 03:27  failure
2026-09-06 03:28  failure
2026-09-07 03:30  failure

Every run fails with the same error, visible in our telemetry:

Contract execution would revert during startingPrice on POP_RULES
  contract: 0x747B456bE03aec0b42bd85C51513730FBD45DA31

Why

DotNS was redeployed on paseo-next-v2 on 2026-09-01 — the day your last green run happened. The new deployment removed the PopRules.startingPrice() function and replaced it with a per-label price() plus a versioned pricing tuple.

package.json pins:

"dependencies": { "bulletin-deploy": "0.15.0" }

0.15.0 calls startingPrice() unconditionally. That function no longer exists on the contract, so the call reverts before anything is deployed.

This is not specific to your repo — it broke every consumer on paseo-next-v2 at the same moment. We shipped a fix in 0.16.1 (2026-09-02) that detects which DotNS generation an environment runs and routes through the matching adapter, so a single version now works against both.

The fix

-  "bulletin-deploy": "0.15.0"
+  "bulletin-deploy": "0.17.0"

Current @latest is 0.17.0 (released 2026-09-04). Nothing else should need to change — your workflow's deploy invocation is unaffected.

Three behaviour notes in 0.16.x → 0.17.0 that are unlikely to matter here but worth a glance:

• isAutoReauthorizeAllowed became an allowlist requiring network: "testnet" explicitly. Only affects library callers passing bulletinAutoAuthorize.
• Non-interactive phone-confirmation failures now exit 78 rather than 1. Only reachable with a session-based signer, not your pool-fallback path.
• Your first deploy after upgrading may re-upload slightly more than usual, once. A filename-hash classification bug meant some bundled assets were re-uploaded on every deploy; fixing it shifts chunk boundaries a single time, then they settle. Subsequent deploys upload less than before.

One suggestion, offered lightly

Your workflow comments say the pin is deliberate — reproducibility over "latest" — which is a completely reasonable default and we do the same in places.

The wrinkle is that bulletin-deploy talks to a live chain, and the chain moves independently of your pin. A pinned client plus a moving chain means a pin is not actually reproducible over time: it reproduces the client, while the thing it talks to changes underneath. That is precisely what happened here — the pin held perfectly and the environment moved.

Options, in increasing order of effort:

• Pin to a caret range (^0.17.0) so patch and minor fixes arrive automatically. Chain-compatibility fixes have consistently been patch or minor for us.
• Keep the exact pin, but treat a red nightly as actionable rather than ambient — six consecutive failures went unnoticed, which suggests the signal is not reaching anyone.
• Call our reusable workflow (paritytech/bulletin-deploy/.github/workflows/deploy.yml@<tag>) instead of invoking the CLI directly, so a bump is one tag change and you inherit fixes without tracking our internals.

Happy to help with any of these, or to answer questions. Filed from a telemetry sweep on our side, not from anything you did wrong.

Start in package.json and inspect the E2E Chain Validation workflow's deploy invocation. Change the bulletin-deploy pin from 0.15.0 to 0.17.0, then rerun the workflow and confirm that chain validation completes successfully without the startingPrice revert.