Cloudflare Logpush → R2 Setup Guide

Cloudflare Logpush → R2 Setup Guide

Long-term log retention for bloqr-backend via Cloudflare Logpush to R2.

---

Why Logpush?

Cloudflare Workers Logs have a 24-hour TTL — logs older than one day are automatically discarded. Logpush delivers a continuous stream of Worker trace events and structured logs to an R2 bucket (or any supported destination), giving you indefinite retention at low cost.

---

Prerequisites

1. Cloudflare account with Workers Logs and Logpush enabled (included in Workers Paid plans).
2. An R2 bucket for log storage — a dedicated bucket is recommended to keep logs separate from filter-list storage:

   # Create a dedicated R2 bucket for logs (recommended)
   wrangler r2 bucket create bloqr-backend-logs
   
   # Alternatively, re-use the existing FILTER_STORAGE bucket (not recommended —
   # mix of application data and logs makes queries harder)

3. logpush = true is already set in wrangler.toml — this flag enables Logpush eligibility for the Worker. The commands below create the actual delivery pipelines.

---

Setup commands

Worker trace requests (errors and non-ok outcomes)

Delivers a filtered stream of Worker invocation traces — useful for debugging failures without storing every successful request:

wrangler logpush create \
  --dataset workers-trace-requests \
  --destination-conf "r2://bloqr-backend-logs/{DATE}/trace-requests" \
  --filter '{"where":{"key":"Outcome","operator":"!eq","value":"ok"}}' \
  --fields "Outcome,ScriptName,EventTimestampMs,Logs,Exceptions,RequestUrl,RequestMethod,ResponseStatus"

All Worker logs (structured JSON logs)

Delivers every structured log statement emitted by the Worker (console.log, AdminLogger, AnalyticsService):

wrangler logpush create \
  --dataset workers-logs \
  --destination-conf "r2://bloqr-backend-logs/{DATE}/worker-logs"

Manage Logpush jobs

# List all configured Logpush jobs
wrangler logpush list

# Inspect a specific job
wrangler logpush describe <job-id>

# Pause a job temporarily
wrangler logpush update <job-id> --enabled false

# Delete a job
wrangler logpush delete <job-id>

---

Querying logs from R2

Logs land in R2 as NDJSON (newline-delimited JSON), partitioned by date:

bloqr-backend-logs/
  2026-03-15/
    trace-requests/   ← filtered Worker trace events
    worker-logs/      ← all structured log lines

Fetch and query with wrangler + jq

# List objects for a given date
wrangler r2 object list bloqr-backend-logs --prefix "2026-03-15/"

# Download and query errors
wrangler r2 object get bloqr-backend-logs "2026-03-15/trace-requests/..." \
  | jq 'select(.Outcome != "ok") | {ts: .EventTimestampMs, url: .RequestUrl, status: .ResponseStatus}'

Forward to Axiom or Grafana Loki

Add a second Logpush job targeting Axiom’s HTTP endpoint or a Loki push API:

# Example: forward to Axiom (replace with your actual endpoint and token)
wrangler logpush create \
  --dataset workers-logs \
  --destination-conf "https://api.axiom.co/v1/datasets/<dataset>/ingest?token=<api-token>"

---

Cost

Resource	Rate
R2 storage	~$0.015 / GB / month
R2 operations	$0.36 / million Class B reads
Logpush	No additional compute charge

Typical Worker log volume is well under 1 GB/month for moderate traffic — total cost is usually under $0.05/month.

---

Notes

• logpush = true in wrangler.toml only marks the Worker as eligible for Logpush. The jobs above must be created separately via wrangler logpush create or the Cloudflare dashboard.
• Each Logpush job is independent — you can have one for trace events and another for structured logs, each with different retention or destination settings.
• R2 bucket names must be globally unique within your Cloudflare account.