IL-0564 Linode · 2025-09-14networking

Linode's Miami, Milan and Sao Paulo data centers had packet loss as traffic moved onto congested alternate paths

The routing between Akamai's core router and edge router has a limit on the number of permitted prefixes. A gradual increase in prefixes reached that limit, the safety limit was triggered, and traffic from the data centers moved to alternate paths. The alternate paths became congested, which caused packet loss in Miami and Milan on 14 September and in Sao Paulo on 15 September. No ISP problem was found that contributed. Akamai raised the prefix limit on all the data center routers to end the impact, and said it would raise the thresholds of the sessions between its backbone and the data center ECOR routers and improve its alerting before limits are reached.

Severity: –
Scope: packet loss and performance degradation for services in the US-MIA, IT-MIL and BR-GRU data centers
Users affected: –
Region: US-MIA (Miami), IT-MIL (Milan), BR-GRU (Sao Paulo)
Time to detect: –
Time to resolve: 10h 40m

Symptoms

Customers using Linode's US-MIA (Miami) and IT-MIL (Milan) data centers saw packet loss and brief performance problems on 14 September 2025 from 17:42 to 18:43 UTC. BR-GRU (Sao Paulo) had the same packet loss on 15 September 2025 from 01:44 to 04:22 UTC.

The status page warned that users might see high packet loss for all services deployed in the affected data centers. How much a customer noticed depended on the products and services they were connected to, and most saw performance degradation.

Miami and Milan were confirmed clear at 20:10 UTC on 14 September and kept under monitoring. The problem then appeared in Sao Paulo from 01:44 UTC on 15 September, while Miami and Milan stayed operational.

Symptom Packet loss

Root cause

Root cause Capacity exhaustion

Component BGP routing Backbone network (WAN)

Contributing factor –

Trigger: A gradual increase in prefixes pushed the routing between the core router and the edge router past the maximum number of permitted prefixes.

Detection

–

Remediation

Raised the prefix limit on all the data center routers

Action items

Raise the thresholds of the sessions between Akamai's backbone and the data center ECOR routers
Improve the existing alerting and check its severity before limits are reached, with escalation to the right team for immediate action

Lessons

–

Sources

https://status.linode.com/incidents/kbvg7325nc3b · official-status · archived copy

Related cases

IL-0350 Hosts in an IONOS TXL cluster lost connectivity when an InfiniBand fabric resweep failed on a full multicast tableIONOS Cloud · 2026
same root cause · shared BGP routing
IL-0341 Traffic between Cloudflare and AWS us-east-1 had high latency, packet loss and origin failures for almost four hoursCloudflare · 2025
same root cause · shared BGP routing

Check your cluster for this

kubegap audits a Kubernetes cluster for conditions like the ones in this case, BGP routing, Backbone network (WAN) among them, read-only, from Claude Code or Codex. Its checks cite public reports like this one.

Open kubegap

For AI tools and scripts

This case as JSON: https://incidentlib-api.benchlet.com/cases/IL-0564
The cases most like it: https://incidentlib-api.benchlet.com/cases/IL-0564/similar
MCP server (Streamable HTTP, no key): https://incidentlib-mcp.benchlet.com/mcp

The API needs no key. The MCP server offers search_incidents, get_incident, list_incidents, get_topic and vocab; both are explained on API and MCP.