Hands-On Microservices with Kubernetes

Handling failures

Istio provides many mechanisms to deal with failure, including the following:

• Timeouts
• Retries (including backoff and jitter)
• Rate limiting
• Health checks
• Circuit breakers

All of these can be configured through Istio CRDs.

For example, the following code demonstrates how to set the connection limits and timeout for the link-manager service at the TCP level (HTTP is supported too):

apiVersion: networking.istio.io/v1alpha3kind: DestinationRulemetadata:   name: link-managerspec:  host: link-manager  trafficPolicy:     connectionPool:       tcp:         maxConnections: 200         connectTimeout: 45ms         tcpKeepalive:           time: 3600s           interval: 75s

Circuit breaking is done by explicitly checking for application errors (for example, the 5XX HTTP status code) within ...