Update: this issue has been ongoing ~10 hours. Other R2 buckets are fine — only this one bucket/prefix is affected.
Key symptom: TTFB is normal (~0.11–0.18s) but throughput is intermittently extremely slow (21 KB/s – 1.2 KB/s), then returns to normal (2.6–4.5 MB/s) for the SAME file from the same client/POP (YYZ). Speed difference up to 200x.
Example slow Ray IDs:
- 9bf2d3c63d7ecbfd-YYZ (total 187s, 1.2 KB/s)
- 9bf2f5467e01076d-YYZ (timeout 60s, 21 KB/s)
- 9bf2f6c38ad6cca0-YYZ (timeout 60s, 33 KB/s)
Example normal Ray IDs:
- 9bf2f8d76e88ab21-YYZ (0.87s, 2.6 MB/s)
- 9bf2f8e2fa5c3a53-YYZ (0.50s, 4.5 MB/s)
Could a Cloudflare engineer check for intermittent routing/throttling/backend degradation affecting this bucket/prefix in ENAM?
I can provide headers (cf-cache-status/range) if needed.