REST API /repos/{owner}/{repo}/actions/runs/{run_id}/attempts/{attempt_number}/logs returns 404 while workflow run is still running

[cunla]

Select Topic Area

Bug

Body

It is impossible to receive the logs of a workflow-run that is in-progress because of that reason.

Also using https://api.github.com/repos/OWNER/REPO/actions/jobs/JOB_ID/logs for getting a specific job logs it returns 404 while the job is running.

[tyteen4a03]

+1

[officalcom]

+1

[fladx]

+2

[JuCebir]

+3

[dfoulkes]

+1

[kamushadenes]

+1

[alaleau]

👍

[NikitaPersefoni]

+1

[SoumithChala]

+1

[lucagonzalez]

+1

[thomassvensen]

+1

[dapzer]

+1

[WaelOuni]

+1

[CAPTAINJAYKIRK]

YES

[Ayfri]

Any update ?

[officalcom]

hallo

[officalcom]

+1

[jmkim0117]

+1

[jonathansant]

+20k

[ajayzeller]

+1

[mykola-digtiar]

+1

[vikash-paf]

+1

[wnmartens]

+1

[s-abdullaev]

+1

[jfglzs]

+1

[AjaysDevWorkspace]

Yeah, this is expected behavior. The logs API usually returns a 404 while the workflow or job is still running — logs are only available once the run is complete.

Would be great if GitHub supported partial or live logs, but as of now, we just have to wait for it to finish.

[mhattingpete]

This is a timing issue with the GitHub Actions API. The logs endpoint returns 404 because logs aren't immediately available during workflow execution. Here's the solution:

Root Cause

• Logs are streamed during execution, not stored until job completion
• The /logs endpoint only works for completed workflow runs
• During execution, you need the real-time streaming approach

Solution 1: Wait for Completion

# Poll until workflow completes
while true; do
  status=$(curl -s -H "Authorization: token $TOKEN" \
    "https://api.github.com/repos/$OWNER/$REPO/actions/runs/$RUN_ID" | \
    jq -r '.status')

  if [[ "$status" == "completed" ]]; then
    # Now logs are available
    curl -H "Authorization: token $TOKEN" \
      "https://api.github.com/repos/$OWNER/$REPO/actions/runs/$RUN_ID/attempts/1/logs"
    break
  fi

  echo "Status: $status, waiting..."
  sleep 30
done

Solution 2: Use Job-Level Logs (More Reliable)

Get individual job logs instead of run logs

# 1. Get jobs for the run
curl -H "Authorization: token $TOKEN" \
  "https://api.github.com/repos/$OWNER/$REPO/actions/runs/$RUN_ID/jobs" | \
  jq -r '.jobs[].id'

# 2. Get logs for specific job (works during execution)
curl -H "Authorization: token $TOKEN" \
  "https://api.github.com/repos/$OWNER/$REPO/actions/jobs/$JOB_ID/logs"

Solution 3: Real-Time Streaming (Advanced)

import requests
import time

def stream_workflow_logs(owner, repo, run_id, token):
    headers = {"Authorization": f"token {token}"}

    while True:
        # Check run status
        run_response = requests.get(
            f"https://api.github.com/repos/{owner}/{repo}/actions/runs/{run_id}",
            headers=headers
        )

        if run_response.json()["status"] == "completed":
            # Get final logs
            logs_response = requests.get(
                f"https://api.github.com/repos/{owner}/{repo}/actions/runs/{run_id}/logs",
                headers=headers
            )
            return logs_response.content

        # Get job logs during execution
        jobs_response = requests.get(
            f"https://api.github.com/repos/{owner}/{repo}/actions/runs/{run_id}/jobs",
            headers=headers
        )

        for job in jobs_response.json()["jobs"]:
            if job["status"] in ["in_progress", "completed"]:
                job_logs = requests.get(
                    f"https://api.github.com/repos/{owner}/{repo}/actions/jobs/{job['id']}/logs",
                    headers=headers
                )
                if job_logs.status_code == 200:
                    print(f"Job {job['name']} logs available")

        time.sleep(10)

Solution 4: GitHub CLI Alternative

# Real-time log following
gh run view $RUN_ID --log

# Or watch specific job
gh run view $RUN_ID --job=$JOB_ID --log

API Behavior Summary

• During execution: Use /jobs/{job_id}/logs endpoint
• After completion: Use /runs/{run_id}/logs endpoint
• Real-time: Poll job logs every 10-30 seconds

Workaround for Your Use Case

# Robust script that handles the timing issue
get_workflow_logs() {
  local run_id=$1
  local max_attempts=60  # 30 minutes max wait
  local attempt=0

  while [ $attempt -lt $max_attempts ]; do
    # Try the run logs endpoint
    if curl -sf -H "Authorization: token $TOKEN" \
        "https://api.github.com/repos/$OWNER/$REPO/actions/runs/$run_id/logs" \
        -o "logs_$run_id.zip"; then
      echo "Logs downloaded successfully"
      return 0
    fi

    # Fall back to job logs if run logs not ready
    curl -s -H "Authorization: token $TOKEN" \
      "https://api.github.com/repos/$OWNER/$REPO/actions/runs/$run_id/jobs" | \
      jq -r '.jobs[].id' | while read job_id; do
        curl -s -H "Authorization: token $TOKEN" \
          "https://api.github.com/repos/$OWNER/$REPO/actions/jobs/$job_id/logs" \
          >> "partial_logs_$run_id.txt"
    done

    sleep 30
    ((attempt++))
  done
}

The 404 is expected behavior - you need to handle the timing appropriately.

[okas]

+1

[josealdaco]

+1

[glauber-shibata]

+1

[marcusvoltolim]

UP