NVIDIA / NVIDIA/nvcf

worker: add metrics for QUIC transport rotation outcomes

Open
#1,384 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Go
Stars
218
Forks
72
Avg merge
1d 12h
Merged PRs (30d)
427

Description

Description

Add an operational metric for worker QUIC transport rotation attempts. PR #1383 adds self-healing transport rotation, but successful recovery and replacement-socket bind failures are currently visible only in logs.

The metric should let operators distinguish a recovered stale network path from a worker that cannot create a replacement socket. A sustained successful rotation rate can also indicate an underlying network or load-balancer problem.

Definition of Done

  • Add a counter for QUIC transport rotation attempts with a bounded result label such as success and bind_error.
  • Define success as a replacement UDP socket being bound and installed for new dials.
  • Pre-initialize every supported result value so the series is present before the first event.
  • Do not use hostnames, addresses, function IDs, or request IDs as metric labels.
  • Add focused tests for successful and failed rotation accounting.
  • Document the intended operational use: successful rotations are diagnostic, while repeated bind failures require operator attention.

Resources (optional)

  • #1383

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the worker QUIC transport rotation implementation introduced by PR #1383 and trace where replacement sockets are bound and installed. Add focused tests for successful and failed rotation accounting, pre-initialize the bounded result values, and document that successful rotations are diagnostic while repeated bind failures need operator attention.

Written by the indexing model from the issue text.

Assessment

Tech stack
go
Domain
networking, observability
Issue type
Feature
Difficulty
3/5
Estimated time
1-2 days
Activity status
Active
Clarity
Clearly specified
Newbie friendliness
74/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.