Executive Summary On February 24, 2026, we experienced a brief partial outage affecting Load Balancers in the BOM1 and BOM2 regions. During this period, load balancers were unable to correctly process Layer 7 (L7) routing policies, which may have resulted in degraded performance or misrouted traffic for services utilizing these specific rules. The issue was traced to an unexpected software interaction between the security policy engine and the L7 rule processor. A patch was swiftly developed and deployed, restoring full functionality.
Impact
Affected Services: Load Balancers (specifically L7 routing policies).
Affected Regions: India (BOM1) and India (BOM2).
Customer Impact: Customers relying on routing rules (such as path-based or host-based routing) behind the affected load balancers may have experienced misrouted requests, 5xx errors, or dropped connections. Standard L4 routing and load balancers without active L7/security policies were not impacted.
Root Cause The disruption was caused by an unhandled error condition within the load balancer software. Specifically, an unexpected interaction occurred between the newly updated security policy engine (WAF) and the L7 rule processing module. When evaluating specific inbound traffic patterns against the security engine, an unhandled exception caused the L7 policy execution to halt, preventing the load balancer from correctly routing the affected requests.
Remediation Upon identifying the unhandled exception, our engineering team immediately developed a software patch to gracefully handle the error condition and ensure the L7 rule processor could continue executing regardless of the security policy engine's interaction state. This patch was safely rolled out to the control planes in both BOM1 and BOM2, instantly resolving the routing failures.
Preventative Measures To prevent this issue from recurring, we are taking the following steps:
Enhanced Test Coverage: We are expanding our automated CI/CD integration testing suites to specifically simulate complex interactions between the security policy engine and L7 routing rules before deployment.
Granular Error Alerting: We are implementing deeper telemetry within the load balancer processing pipeline to immediately flag unhandled exceptions in policy evaluation, further reducing time-to-detection for similar edge cases.
Graceful Failover: We are reviewing the load balancer architecture to ensure that non-critical sub-system faults (like specific security policy evaluations) do not block core routing functions.
