Telemetry submit endpoint rejects handshake: telemetry.quantus.cat/submit/ returns 502/404 on every connect

[adamtpang]

Environment

• Node version: 0.6.4-phase-alignment-373ecfe4f9e
• Binary: quantus-node-v0.6.4-phase-alignment-x86_64-pc-windows-msvc.zip
• OS: Windows 11 Home, 10.0.26200, x64
• Chain: planck
• Telemetry URL (as logged by node): wss://telemetry.quantus.cat/submit/

Symptom

Every telemetry handshake attempt from the node is rejected at the WSS upgrade with HTTP 502 or 404. Over a ~42-minute session on planck:

Metric	Count
Handshake rejected (502)	58
Handshake rejected (404)	3
Successful connections (log-observable)	0

Representative log lines:

2026-04-21 18:34:28 ❌ Error while dialing /dns/telemetry.quantus.cat/tcp/443/x-parity-wss/%2Fsubmit%2F:
  Custom { kind: Other, error: Other(Left(Handshake("server rejected handshake; status code = 502"))) }

2026-04-21 18:35:40 ❌ Error while dialing /dns/telemetry.quantus.cat/tcp/443/x-parity-wss/%2Fsubmit%2F:
  Custom { kind: Other, error: Other(Left(Handshake("server rejected handshake; status code = 404"))) }

First and last errors are both 502, ~42 minutes apart. Not a transient blip.

Reproduction via curl

Confirms server-side behaviour is independent of the node client:

$ for path in /submit /submit/ /submit/v1 /feed/; do
    printf "%s -> " "$path"
    curl -s -o /dev/null -w "%{http_code}\n" "https://telemetry.quantus.cat$path" --max-time 5
  done
/submit   -> 404
/submit/  -> 404
/submit/v1 -> 404
/feed/    -> 404

(At different times during the session, /submit/ also returned 502.)

Note: node still appears on telemetry UI

Despite 0 log-observable successful connections, the node does appear on the telemetry UI (telemetry.quantus.cat, Planck tab) with the correct name, version, peer count, and chain height updating in near-real-time. This suggests either:

• A prior connection at a different codepath/endpoint is feeding the UI
• The frontend is reading cached/different data than what /submit/ accepts
• Handshake rejections are happening after the upgrade that the UI cares about

Would help to know which is authoritative.

Suggested investigation

1. Confirm whether wss://telemetry.quantus.cat/submit/ is the intended submission endpoint for planck. The chain-spec-embedded URL and any documented override should agree.
2. Check telemetry ingester health / nginx routing to /submit/ (the 502/404 split smells like upstream-down + routing-not-configured for some paths).
3. Consider surfacing in node docs / release notes what the currently correct telemetry URL is per chain, so operators don't spend time guessing between telemetry.quantus.cat and shard-telemetry.quantus.cat.

Non-blocking for mining

Confirmed separately that mining itself is unaffected: block import, peer mesh, and authorship all work normally. RPC (system_health, chain_getHeader) is the authoritative local check. Filing this as a telemetry/infra bug, not a node bug.

Happy to share full redacted logs or test other endpoint candidates if useful.