Detect invalid configurations

Stay organized with collections Save and categorize content based on your preferences.

Network Analyzer lets you detect configurations that are not working as expected. These invalid configurations can be due to some setup errors or regressions caused by other changes. When different teams own the services affected by such invalid configuration, it can be hard to troubleshoot such issues. Discovering such failures as they happen and identifying the root causes can help in faster troubleshooting and effective communication between different teams.

Configuration errors due to blocking firewalls

Firewall configuration errors that block Google Cloud services can happen during or after the initial setup.

During initial setup

Some of the common firewall errors that block Google Cloud services from functioning are as follows:

  • Missing firewall configuration. For example, the load balancer's health check firewall has not been configured.
  • Typos in the manual configuration process that cause faulty configurations.
  • Inconsistent configuration caused by missing VM tags. For example, you configure the firewall rule with the expected target VM tags, but some of the backend VMs have not been associated with the specific tag.

After the initial setup

An unintended firewall configuration change can break a service that has been functioning properly. For example, you can unintentionally create a higher priority firewall deny rule that blocks connectivity to GKE or Cloud SQL services.

Scenario: Health check firewall is not configured for the load balancer

In this example, Network Analyzer reports an insight in the category of load balancer insights of type health check firewall is not configured. The insight details page shows the implied deny ingress rule in the network where the load balancer is configured. The deny ingress rule blocks the health check range. This indicates that the load balancer's backends do not have firewall rules configured to allow the health check range.

Configure the health check firewall rule to explicitly allow the health check range to access the load balancer backends.

Scenario: GKE node to control plane connectivity blocked by a firewall rule

In this example, an insight that belongs to the category of GKE node connectivity insights of type GKE node to control plane connectivity is generated.

The insight details page shows that a firewall rule is denying the connection from the GKE node to the control plane in a cluster. This issue is because of a denying rule; remove this rule or set up a higher priority allowing rule to address the issue.

Scenario: GKE control plane to node connectivity blocked by a firewall rule

In this example, an insight that belongs to the category of GKE control plane connectivity insights of type GKE control plane to node connectivity is generated.

This insight details page shows that a firewall rule is denying the connection from the GKE control plane to its nodes in a cluster.

Routing configuration errors

Routes with invalid next hops can cause partial or total traffic loss. Such loss can be due to having routes to next hop VMs that are not running or have been deleted.

Configuration changes that can cause unintended routing change include the following scenarios:

  • Adding a new subnet with IP address ranges that overlap with a dynamic route, results in shadowing the dynamic route and this might lead to a traffic drop.
  • Adding a new default route through a VPN can cause capacity bottlenecks that affect network performance.

Scenario: The VM in the next hop is deleted

In this example, an insight from the category of routes with invalid next hops with the VM is deleted type appears.

The insight details page shows that the next hop VM in this route is deleted, so this route is reported as invalid.

Either delete the route or create a new VM instance to be used as the next hop of the route. A configuration change that might cause this insight is displayed in the insight details page. Click the link to go to the Cloud Logging page to see the details of the configuration such as the user who made the change and the time of the change.

Scenario: The dynamic route is shadowed

In this example, an insight from the shadowed dynamic routes insights category of fully shadowed by peering subnet route type is generated.

This insight details page shows that an imported dynamic route learned from the network peer with next hop peering network dynamic-routes is being shadowed. The IP address range of the dynamic route overlaps with a new subnet route and thus it is fully shadowed. Traffic that goes to the peering network to this IP address range is forwarded to a subnet in the VPC network.

Scenario: Connectivity to Cloud SQL instance blocked by missing network peering

In this example, an insight from the Cloud SQL connectivity insights category of connectivity to Cloud SQL instance blocked by routing issue type is displayed.

The insight page shows that the connectivity to a Cloud SQL instance is blocked because the network peering is missing.