惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

J
Java Code Geeks
F
Fortinet All Blogs
云风的 BLOG
云风的 BLOG
MyScale Blog
MyScale Blog
D
DataBreaches.Net
Stack Overflow Blog
Stack Overflow Blog
A
About on SuperTechFans
Google DeepMind News
Google DeepMind News
Microsoft Security Blog
Microsoft Security Blog
腾讯CDC
The GitHub Blog
The GitHub Blog
Jina AI
Jina AI
B
Blog RSS Feed
I
InfoQ
N
Netflix TechBlog - Medium
T
The Blog of Author Tim Ferriss
Microsoft Azure Blog
Microsoft Azure Blog
Recent Announcements
Recent Announcements
GbyAI
GbyAI
H
Help Net Security
L
LangChain Blog
M
MIT News - Artificial intelligence
Y
Y Combinator Blog
aimingoo的专栏
aimingoo的专栏

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
Why Your GitHub Actions Workflow Didn't Trigger: branches...
opscanopy · 2026-06-14 · via DEV Community

You pushed a commit, opened the Actions tab, and there's nothing there. No red X, no yellow dot — the workflow simply didn't run. There's no error to read, no log to grep, because a workflow that doesn't trigger produces no run at all. The decision happened before any runner was assigned, inside GitHub's event-filtering logic, and that logic is more surprising than the docs make it look.

Almost every "why did my GitHub Actions workflow not trigger" report comes down to one of a handful of causes: the workflow file isn't on the branch you pushed to, your branches filter doesn't match the ref, or — the big one — you combined branches and paths without realizing they're ANDed together. Here's each cause with the deciding rule and the fix.

1. The workflow file isn't on the target branch

GitHub reads on: triggers from the version of the workflow file that exists on the branch receiving the event — not from your default branch. If you added .github/workflows/ci.yml on main but push to a feature/x branch that branched off before that file existed, there's no workflow to trigger there.

# on main, but feature/x branched before this file existed
on:
  push:
    branches: ['**']

This is the most common false alarm. The fix is mechanical: merge or rebase main into the branch so the workflow file is present, then push again. The same rule explains why edits to on: triggers only "take effect" once the change reaches the branch you're testing on.

Why it matters: there is no error message for "no workflow file here." It's the first thing to rule out before you suspect your filters.

2. The branch filter doesn't match the ref

branches and tags are glob patterns, and the glob rules are stricter than shell globs. A plain * matches one path segment — it stops at /. To match across slashes you need **.

# BAD — '*' does not cross '/', so 'release/1.2' never matches
on:
  push:
    branches:
      - 'release/*'   # matches release/1.2 ... actually this IS fine
      - 'feature*'    # matches 'feature' and 'featureX' but NOT 'feature/login'

The trap is feature* versus feature/**. feature* matches the literal segment featureX, but a branch named feature/login contains a slash, and * won't cross it. You want feature/**.

# FIXED — ** crosses slashes
on:
  push:
    branches:
      - 'release/**'
      - 'feature/**'
      - main

The glob characters GitHub honors: * (any chars except /), ** (any chars including /), ? (one char), + (one or more of the preceding), [] character ranges, ! at the start of a pattern to negate, and \ to escape a special character (so \* matches a literal asterisk). Order matters for negation — a later !pattern excludes refs an earlier pattern included.

Why it matters: * not crossing / is responsible for a huge share of "github actions branches filter not working" reports. When in doubt, reach for **.

3. The AND-semantics of branches + paths

This is the one that burns experienced engineers. When a push or pull_request event has both a branch filter and a path filter, the event must satisfy both to trigger. They are ANDed, not ORed.

# BAD — intent: "run on a push to main, OR when src changes"
# reality: "run only on a push to main AND when src/** changed"
on:
  push:
    branches: [main]
    paths: ['src/**']

A push to main that only touches README.md will not run this workflow — the branch matched, but no path did, and both must hold. People read this block as an OR and are baffled when docs-only commits skip CI.

If you genuinely want "main pushes always, plus any branch when src changes," that's two separate filter sets, which on: can't express in one push block — you split it across triggers or use job-level if: conditions on github.ref instead.

# FIXED — be explicit that you want both conditions, or drop one
on:
  push:
    branches: [main]
    paths:
      - 'src/**'
      - '.github/workflows/**'   # so CI changes still trigger

Why it matters: the AND-semantics are documented in one sentence and contradict most people's intuition. If your workflow "randomly" skips some pushes to the right branch, a path filter is almost always the cause.

4. paths with no branches companion still needs a real ref

A subtle corollary: when you filter on.push.paths and want it to apply across all branches, you don't need a branches block at all — omitting it means "all branches." But the moment you add branches, rule #3 kicks in. People sometimes add branches: ['**'] thinking it's required for paths to work; it isn't, and adding it changes nothing because ** matches every branch anyway. The thing to internalize is that a missing filter means "match everything," and a present filter narrows.

# These behave identically: paths applies to every branch
on:
  push:
    paths: ['src/**']
# vs
on:
  push:
    branches: ['**']
    paths: ['src/**']

5. paths-ignore and the diff that's too big

paths-ignore skips the run only if every changed file matches an ignore pattern. If a single file falls outside the ignore list, the workflow runs. So one stray change defeats the whole filter — which is usually what you want, but surprises people who expect "ignore these files" to mean "ignore commits that touch these files."

# Skips ONLY when every changed file is docs; one code file => runs
on:
  push:
    paths-ignore:
      - 'docs/**'
      - '**.md'

Two more gotchas live here. First, path filters are evaluated against the diff, and GitHub only inspects up to 300 changed files (1,000 commits) — beyond that limit, path filtering gives up and the workflow runs (or is evaluated as if the filter passed). A giant force-push or a huge merge can trigger a workflow your paths-ignore "should" have skipped. Second, you cannot mix paths and paths-ignore in the same trigger; pick one.

Why it matters: paths-ignore is an all-or-nothing gate on the diff, and the 300-file ceiling means it's not a hard guarantee on large changes.

6. pull_request, forks, and pull_request_target

Branch filters on pull_request match the base branch (where the PR will merge), not the head branch the contributor is working on. If you write branches: [main] expecting it to match the contributor's feature/x, it won't — it matches PRs targeting main.

# Runs on PRs whose BASE (merge target) is main or a release branch
on:
  pull_request:
    branches:
      - main
      - 'release/**'

And pull_request from a fork is restricted: a first-time contributor's PR may require manual approval before any workflow runs, which looks identical to "didn't trigger." If you switched to pull_request_target to get around fork restrictions, note that it reads the workflow and triggers from the base branch's version of the file — and carries real security risk, covered in our GitHub Actions security misconfigurations post.

A copy-paste filter cheat-sheet

on:
  push:
    branches:                 # ref globs; missing = all branches
      - main
      - 'release/**'          # ** crosses '/'; '*' does not
      - 'feature/**'
    tags:
      - 'v[0-9]+.[0-9]+.[0-9]+'   # numeric semver tags only
    paths:                    # ANDed with branches — BOTH must match
      - 'src/**'
      - '.github/workflows/**'
  pull_request:
    branches: [main]          # matches the PR's BASE branch
    paths-ignore:             # skip only if EVERY changed file matches
      - '**.md'

Quick reference for the glob characters: * = any chars except /, ** = any chars including /, ? = one char, + = one-or-more of the preceding, [a-z] = range, leading ! = negate, \ = escape.

Stop guessing — replay your event

The reason these bugs are maddening is that the feedback loop is "push and pray." There's no dry-run, no --explain, just an empty Actions tab. So you commit a one-line change, push, refresh, wait, and repeat — burning minutes per guess against semantics you're not sure of.

The GitHub Actions Expression & Trigger Tester closes that loop. Paste your on: block, describe the event — push to feature/login, tag v2.1.0, or a pull_request targeting main with a list of changed files — and it evaluates every branches, tags, paths, and paths-ignore filter with the same glob engine and AND-semantics GitHub uses. You get a per-job RUNS / SKIPPED table with the exact deciding reason: "branch matched, but no path filter did," or "* does not cross /." It's 100% in your browser — your workflow YAML never leaves the page.

See exactly which jobs run before you push, not after.

Open the GitHub Actions Expression & Trigger Tester →

Originally published on OpsCanopy. Try it free, in your browser: CVE-Ignore Converter.