# Bunny.net: Isolated 502s for Magic Containers

- Impact (vendor’s label): major
- Status: resolved
- Started: 2026-04-30 10:12 UTC · Resolved: 2026-05-26 08:23 UTC · Duration: 26 d
- Updates: 6 · Affected components: Magic Containers, Edge Scripting, Bunny Database
- [Vendor status page: https://stspg.io/2dvsbxdh2t76](https://stspg.io/2dvsbxdh2t76)
- [Company: Bunny.net](https://rewindex.com/company/bunny-net.md)

## Summary

Major isolated 502s affected North American CDN endpoints using Magic Containers and Databases, starting 2026-04-30 10:12 UTC and resolved 2026-05-26 08:23 UTC (25 days 22 h). Database routing adjustments restored service; Bunny.net later fixed delayed endpoint synchronization between Magic Containers and CDN routing, linked to slow sticky session invalidation that left stale endpoint data in use. _(machine-written, sol)_
## Timeline of updates

- **investigating** 2026-04-30 10:12 UTC: We are currently looking into isolated 502s within the NA region for CDN endpoints utilizing Magic Containers and Databases. This is primarily affecting specific regions in North America. Our Engineering team are investigating the issue currently.
- **investigating** 2026-04-30 10:14 UTC: We are continuing to investigate this issue.
- **investigating** 2026-04-30 12:16 UTC: We have made some routing adjustments to mitigate the timeouts on Database and is fully operational again. Our Engineering team is continuing to investigate the Magic Containers issue.
- **investigating** 2026-05-01 10:48 UTC: Our Engineering team are still investigating the root cause for isolated 502s in the NA region.
- **monitoring** 2026-05-06 14:29 UTC: We are monitoring improvements related to the resilience of routing configuration failover on edge servers during temporary control plane connectivity interruptions.
- **resolved** 2026-05-26 08:23 UTC: We identified and resolved an issue that could cause endpoint updates to experience delayed propagation after CDN services had been running for an extended period of time. The issue was related to endpoint synchronization between Magic Containers and the CDN routing component, where slower processing connected to sticky session invalidation could result in stale endpoint data remaining in use longer than expected.

## Rewindex changes for this incident

- 2026-04-30 10:12 UTC: [Major incident: Isolated 502s for Magic Containers](https://rewindex.com/change/i0a7a1734a923e2.md)


---
Canonical: https://rewindex.com/company/bunny-net/incidents/gpm8dsfrptw1 · Markdown: https://rewindex.com/company/bunny-net/incidents/gpm8dsfrptw1.md · JSON: https://rewindex.com/api/v1/companies/bunny-net/incidents/gpm8dsfrptw1 · Generated 2026-10-18T13:25:49.373Z