Summary
On February 20, 2026, Laravel Cloud experienced a connectivity outage lasting approximately 3 hours and 15 minutes. The disruption was caused by an incident at Cloudflare, one of our infrastructure partners, which resulted in the withdrawal of IP prefix advertisements that route traffic to Laravel Cloud services.
During this time, customers were unable to reach the Laravel Cloud control panel or some of their applications hosted on Laravel Cloud. No customer data was lost or compromised, and no deployments in progress were affected. The issue was entirely network-level — applications and their underlying infrastructure remained healthy throughout.
While redundancy generally exists throughout our infrastructure, we have not yet extended that same protection to our IP announcement layer. Closing that gap is now a top engineering priority. Our goal is to implement additional failover protection so that no single upstream dependency can cause a prolonged outage for our customers.
Timeline (UTC)
| Time | Event |
|---|---|
| 18:42 | Laravel Cloud monitoring detects connectivity failures. Incident declared and response team assembled immediately. |
| 18:45 | Cloudflare begins investigating issues with their services and network. |
| 18:48 | Laravel Cloud status page updated and notification posted to Twitter/X. |
| 19:09 | Cloudflare identifies impact to a subset of BYOIP (Bring Your Own IP) prefixes. Reports the underlying issue as mitigated and begins working to restore affected advertisements. |
| ~19:40 | Laravel Cloud attempts to re-advertise prefixes via the Cloudflare dashboard. The option is unavailable — prefixes are locked in a "Withdrawn" state requiring Cloudflare to unlock. We communicate this to Cloudflare via our shared support channel. |
| 20:50 | Cloudflare acknowledges that some customers are unable to re-advertise their prefixes through the dashboard and begins working on a fix. |
| 21:57 | Full connectivity to Laravel Cloud is restored as Cloudflare completes prefix restoration. |
Total duration of impact: approximately 3 hours and 15 minutes.
What Happened
Laravel Cloud uses Cloudflare for network services including IP prefix management via their BYOIP (Bring Your Own IP) product. On February 20, Cloudflare experienced an incident that caused IP prefix advertisements to be withdrawn for a subset of their customers, including Laravel Cloud.
When IP prefixes are withdrawn from BGP, traffic can no longer be routed to the affected services. This meant that while Laravel Cloud's applications, databases, and infrastructure were fully operational, users could not reach them because the network path no longer existed.
Cloudflare identified the underlying cause and advised customers to re-advertise their prefixes manually through the Cloudflare dashboard. However, Laravel Cloud's prefixes were locked in a "Withdrawn" state that required Cloudflare's intervention to unlock. This meant the suggested self-mitigation path was not available to us, and recovery was dependent on Cloudflare restoring the prefixes on their end.
What Was Affected
- Connectivity to Laravel Cloud applications — users could not reach their apps
- Laravel Cloud dashboard and API — inaccessible during the outage
- Deployments — new deployments could not be initiated
What Was Not Affected
- Customer data — all data remained intact and secure
- Running applications — apps continued to run normally; they were simply unreachable
- Databases and caches — no data loss or corruption
Our Response
Our monitoring detected the issue within minutes, before Cloudflare's own status page reflected the incident. We immediately declared an incident, assembled our response team, and updated our status page and social channels to keep customers informed.
Once Cloudflare published their self-mitigation guidance, we attempted to follow it but were unable to do so due to the prefix locking issue described above. We escalated directly with Cloudflare's team and continued to monitor and communicate throughout the incident.
Looking Forward
Laravel acknowledges that single points of failure in our infrastructure can disproportionately introduce risk in our ability to manage availability of the applications we serve. We are exploring strategies to decouple and introduce redundancy in our key networking layers to create better resiliency from underlying components.
We take the reliability of Laravel Cloud seriously, and we understand the impact this outage had on our customers and their users. We are committed to learning from this incident and building a more resilient platform.