Cloudflare Stream live inputs hit by critical SRT and RTMPS ingest failures
Users are reporting critical service interruptions in Cloudflare Stream's live ingest pipelines using SRT and RTMPS protocols. Despite valid configuration, inputs are returning null status codes and failing to produce recordings, suggesting a possible regression in the platform's ingest or provisioning systems.
Key Takeaways
- SRT handshakes successfully authenticate but fail within seconds due to I/O muxer errors and unacknowledged data packets.
- RTMPS ingest is being rejected at the application layer in under one second, returning 'End of file' errors to encoders.
- API queries for new live inputs show a 'null' status despite configurations being set to automatic recording and enabled.
- Cloudflare's storage API remains frozen at previous levels, indicating a total breakdown in current ingest processing.
- The issue persists across multiple encoder types, including FFmpeg on Windows and mobile implementations using srtdroid.
Why It Matters
This disruption exposes a potential regression in Cloudflare's ingest provisioning, threatening the reliability of its low-latency HLS infrastructure. For developers building on serverless video stacks, such failures undermine the core value proposition of abstracting away ingest complexity. Failure to capture recordings suggests a deeper synchronicity issue between the edge ingest nodes and back-end storage buckets. If Cloudflare continues to struggle with protocol-specific stability, it risks losing market share to competitors like Mux or api.video, which prioritize ingest consistency. Strategists should monitor for an official post-mortem or a fix in the Cloudflare Stream changelog, specifically regarding live input provisioning logic.
Additional Context
The current ingest failures follow a series of infrastructure challenges for Cloudflare in early 2026. Per Cloudflare's own incident reporting from February 2026, the company suffered a major six-hour global outage caused by changes to its Bring Your Own IP (BYOIP) pipeline, which led to unintended BGP route withdrawals. This incident impacted high-traffic platforms including LinkedIn and Zoom, highlighting how iterative configuration changes in its global network can lead to widespread service unreachability. By May 2026, Cloudflare concluded a major resilience initiative dubbed 'Code Orange,' which introduced tools like Snapstone to automate safer configuration deployments. In the video sector specifically, Cloudflare has been aggressively expanding its Stream capabilities to compete with AWS and Mux. Per a BlogGeek.me report from late 2025, Cloudflare acquired Dyte to enhance its client SDK presence and began rolling out managed Media-over-QUIC (MoQ) relays to support next-generation low-latency streaming. While the platform graduated SRT ingest from beta in early 2026 to offer more resilient ingest over lossy networks, the current 'null' status issues suggest that these new protocol paths may still lack the maturity of legacy RTMP stacks. Market competition has intensified as serverless video platforms stratify by reliability and cost. According to an April 2026 analysis by BlazingCDN, Cloudflare Stream’s consumption-based pricing — including the $0.75 per 1,000 minutes live input fee — is targeted at high-volume user-generated content platforms. However, reliability remains a critical differentiator; while Cloudflare offers lower delivery costs than competitors like api.video, ongoing technical regressions in core ingest protocols could shift executive preference toward more specialized video-as-a-service providers that avoid general network infrastructure fluctuations.
Read full article at community.cloudflare.com
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source