Service Resolver Configuration Entry

v1.8.4+: On Kubernetes, the ServiceResolver custom resource is supported in Consul versions 1.8.4+.
v1.6.0+: On other platforms, this config entry is supported in Consul versions 1.6.0+.

The service-resolver config entry kind (ServiceResolver on Kubernetes) controls which service instances should satisfy Connect upstream discovery requests for a given service name.

If no resolver config is defined the chain assumes 100% of traffic goes to the healthy instances of the default service in the current datacenter+namespace and discovery terminates.

Interaction with other Config Entries

Service resolver config entries are a component of L7 Traffic Management.

UI

Once a service-resolver is successfully entered, you can view it in the UI. Service routers, service splitters, and service resolvers can all be viewed by clicking on your service then switching to the routing tab.

screenshot of service resolver in the UI

Sample Config Entries

Filter on service version

Create service subsets based on a version metadata and override the defaults:

Kind          = "service-resolver"
Name          = "web"
DefaultSubset = "v1"
Subsets = {
  v1 = {
    Filter = "Service.Meta.version == v1"
  }
  v2 = {
    Filter = "Service.Meta.version == v2"
  }
}

apiVersion: consul.hashicorp.com/v1alpha1
kind: ServiceResolver
metadata:
  name: web
spec:
  defaultSubset: v1
  subsets:
    v1:
      filter: 'Service.Meta.version == v1'
    v2:
      filter: 'Service.Meta.version == v2'

{
  "Kind": "service-resolver",
  "Name": "web",
  "DefaultSubset": "v1",
  "Subsets": {
    "v1": {
      "Filter": "Service.Meta.version == v1"
    },
    "v2": {
      "Filter": "Service.Meta.version == v2"
    }
  }
}

Other datacenters

Expose a set of services in another datacenter as a virtual service:

Kind = "service-resolver"
Name = "web-dc2"
Redirect {
  Service    = "web"
  Datacenter = "dc2"
}

Failover

Enable failover for subset v2 to dc2, and all other subsets to dc3 or dc4:

Kind           = "service-resolver"
Name           = "web"
ConnectTimeout = "15s"
Failover = {
  v2 = {
    Datacenters = ["dc2"]
  }
  "*" = {
    Datacenters = ["dc3", "dc4"]
  }
}

apiVersion: consul.hashicorp.com/v1alpha1
kind: ServiceResolver
metadata:
  name: web
spec:
  connectTimeout: 15s
  failover:
    v2:
      datacenters: ['dc2']
    '*':
      datacenters: ['dc3', 'dc4']

Enable failover to a WAN federated datacenter and then a cluster peer for all service subsets.

Kind           = "service-resolver"
Name           = "web"
ConnectTimeout = "15s"
Failover = {
  "*" = {
    Targets = [
      {Datacenter = "dc3"},
      {Peer = "peer-01"}
    ]
  }
}

apiVersion: consul.hashicorp.com/v1alpha1
kind: ServiceResolver
metadata:
  name: web
spec:
  connectTimeout: 15s
  failover:
    '*':
      targets:
      - datacenter: "dc3"
      - peer: "peer-01"

{
  "Kind": "service-resolver",
  "Name": "web",
  "ConnectTimeout": "15s",
  "Failover": {
    "*": {
      "Targets": [
        {"Datacenter": "dc3"},
        {"Peer": "peer-01"}
      ]
    }
  }
}

Failover to another datacenter and namespace for all service subsets.

Kind           = "service-resolver"
Name           = "product-api"
Namespace      = "primary"
ConnectTimeout = "0s"
Failover = {
  "*" = {
    Datacenters = ["dc2"]
    Namespace = "secondary"
  }
}

apiVersion: consul.hashicorp.com/v1alpha1
kind: ServiceResolver
metadata:
  name: product-api
  namespace: primary
spec:
  connectTimeout: 0s
  failover:
    namespace: 'secondary'
    datacenters: ['dc2']

{
  "Kind": "service-resolver",
  "Name": "product-api",
  "Namespace": "primary",
  "ConnectTimeout": "0s",
  "Failover": {
    "*": {
      "Datacenters": ["dc2"],
      "Namespace": "secondary"
    }
  }
}

Failover within a datacenter and a different namespace.

Kind           = "service-resolver"
Name           = "product-api"
Namespace      = "primary"
ConnectTimeout = "0s"
Failover = {
  "*" = {
    Service = "product-api-backup"
    Namespace = "secondary"
  }
}

apiVersion: consul.hashicorp.com/v1alpha1
kind: ServiceResolver
metadata:
  name: product-api
  namespace: primary
spec:
  connectTimeout: 0s
  failover:
    service: 'product-api-backup'
    namespace: 'secondary'

{
  "Kind": "service-resolver",
  "Name": "product-api",
  "Namespace": "primary",
  "ConnectTimeout": "0s",
  "Failover": {
    "*": {
      "Service": "product-api-backup",
      "Namespace": "secondary"
    }
  }
}

Consistent load balancing

Apply consistent load balancing for requests based on x-user-id header:

Kind = "service-resolver"
Name = "web"

LoadBalancer = {
  Policy = "maglev"
  HashPolicies = [
    {
      Field = "header"
      FieldValue = "x-user-id"
    }
  ]
}

{
  "Kind": "service-resolver",
  "Name": "web",
  "LoadBalancer": {
    "Policy": "maglev",
    "HashPolicies": [
      {
        "Field": "header",
        "FieldValue": "x-user-id"
      }
    ]
  }
}

Available Fields

Kind - Must be set to service-resolver
Name (string: <required>) - Set to the name of the service being configured.
Namespace (string: "default")Enterprise - Specifies the namespace in which the configuration entry will apply.
Partition (string: "default")Enterprise - Specifies the admin partition in which the configuration entry will apply.
Meta (map<string|string>: nil) - Specifies arbitrary KV metadata pairs. Added in Consul 1.8.4.
ConnectTimeout (duration: 0s) - The timeout for establishing new network connections to this service. The default unit of time is ns.
RequestTimeout (duration: 0s) - The timeout for receiving an HTTP response from this service before the connection is terminated. If unspecified or 0, the default of 15s is used. The default unit of time is ns.
DefaultSubset (string: "") - The subset to use when no explicit subset is requested. If empty, the unnamed subset is used.
Subsets (map[string]ServiceResolverSubset) - A map of subset name to subset definition for all usable named subsets of this service. The map key is the name of the subset and all names must be valid DNS subdomain elements.

This may be empty, in which case only the unnamed default subset will be usable.
- Filter (string: "") - The filter expression to be used for selecting instances of the requested service. If empty, all healthy instances are returned. This expression can filter on the same selectors as the Health API endpoint.
- OnlyPassing (bool: false) - Specifies the behavior of the resolver's health check interpretation. If this is set to false, instances with checks in the passing as well as the warning states will be considered healthy. If this is set to true, only instances with checks in the passing state will be considered healthy.
Redirect (ServiceResolverRedirect: <optional>) - When configured, all attempts to resolve the service this resolver defines will be substituted for the supplied redirect EXCEPT when the redirect has already been applied.

When Redirect is set, all other fields besides Kind, Name, Namespace and Redirect will be ignored.
- Service (string: "") - A service to resolve instead of the current service.
- ServiceSubset (string: "") - A named subset of the given service to resolve instead of one defined as that service's DefaultSubset. If empty, the default subset is used.
  
  If this is specified at least one of Service, Datacenter, or Namespace should be configured.
- Namespace (string: "")Enterprise - Specifies the destination namespace to resolve the service from.
- Partition (string: "")Enterprise - Specifies the destination admin partition to resolve the service from.
- Datacenter (string: "") - Specifies the destination datacenter to resolve the service from.
- Peer (string: "") - Specifies the destination cluster peer to resolve the target service name from. Ensure that intentions are defined on the peered cluster to allow the source service to access this redirect target service as an upstream. When the peer name is specified, Consul uses Envoy's outlier detection to determine the health of the redirect target based on whether communication attempts to the redirect target are generally successful. Service health checks imported from a peer are not considered in the health of a redirect target because they do not account for service routers, splitters, and resolvers that may be defined in the peer for the target service.
Failover (map[string]ServiceResolverFailover) - Controls when and how to reroute traffic to an alternate pool of service instances.

The map is keyed by the service subset it applies to and the special string "*" is a wildcard that applies to any subset not otherwise specified here.

Service, ServiceSubset, Namespace, Targets, and Datacenters cannot all be empty at once.
- Service (string: "") - The service to resolve instead of the default as the failover group of instances during failover.
- ServiceSubset (string: "") - The named subset of the requested service to resolve as the failover group of instances. If empty, the default subset for the requested service is used.
- Namespace (string: "")Enterprise - The namespace to resolve the requested service from to form the failover group of instances. If empty, the current namespace is used.
- Datacenters (array<string>) - A fixed list of datacenters to try during failover.
- Targets (array<ServiceResolverFailoverTarget>) - A fixed list of failover targets to try during failover. It allows operators to express complicated failover scenarios such as between cluster peers, WAN federated datacenters, and local admin partitions.
  - Service (string: "") - The service name to use for the failover target. If empty, the current service name is used.
  - ServiceSubset (string: "") - The named subset to use for the failover target. If empty, the default subset for the requested service name is used.
  - Namespace (string: "")Enterprise - The namespace to use for the failover target. If empty, the default namespace is used.
  - Partition (string: "")Enterprise - The admin partition within the same datacenter to use for the failover target. If empty, the default partition is used. To use an admin partition in a different datacenter for the failover target, use the Peer field instead.
  - Datacenter (string: "") - The WAN federated datacenter to use for the failover target. If empty, the current datacenter is used. To use a datacenter for the failover target that is connected with a cluster peering relationship rather than WAN federation, use the Peer field instead.
  - Peer (string: "") - Specifies the destination cluster peer to resolve the target service name from. Ensure that intentions are defined on the peered cluster to allow the source service to access this failover target service as an upstream. When the peer name is specified, Consul uses Envoy's outlier detection to determine the health of the failover target based on whether communication attempts to the failover target are generally successful. Service health checks imported from a peer are not considered in the health of a failover target because they do not account for service routers, splitters, and resolvers that may be defined in the peer for the target service.
LoadBalancer (LoadBalancer) - Determines the load balancing policy and configuration for services issuing requests to this upstream. This option is available in Consul versions 1.9.0 and newer.
- Policy (string: "") - The load balancing policy used to select a host. One of: random, round_robin, least_request, ring_hash, maglev.
- RingHashConfig (RingHashConfig) - Configuration for the ring_hash policy type.
  - MinimumRingSize (int: 1024) - Determines the minimum number of entries in the hash ring.
  - MaximumRingSize (int: 8192) - Determines the maximum number of entries in the hash ring.
- LeastRequestConfig (LeastRequestConfig) - Configuration for the least_request policy type.
  - ChoiceCount (int: 2) - Determines the number of random healthy hosts from which to select the one with the least requests.
- HashPolicies (array<HashPolicies>) - List of hash policies to use for hashing load balancing algorithms. Hash policies are evaluated individually and combined such that identical lists result in the same hash. If no hash policies are present, or none are successfully evaluated, then a random backend host will be selected.
  - Field (string: "") - The attribute type to hash on. Must be one of header, cookie, or query_parameter. Cannot be specified along with SourceIP.
  - FieldValue (string: "") - The value to hash. ie. header name, cookie name, URL query parameter name. Cannot be specified along with SourceIP.
  - CookieConfig (CookieConfig) - Additional configuration for the "cookie" hash policy type. This is specified to have Envoy generate a cookie for a client on its first request.
    - Session (bool: false) - Generates a session cookie with no expiration.
    - TTL (duration: 0s) - TTL for generated cookies. Cannot be specified for session cookies.
    - Path (string: "") - The path to set for the cookie.
  - SourceIP (bool: false) - Determines whether the hash should be of the source IP address rather than of a field and field value. Cannot be specified along with Field or FieldValue.
  - Terminal (bool: false) - Will short circuit the computation of the hash when multiple hash policies are present. If a hash is computed when a Terminal policy is evaluated, then that hash will be used and subsequent hash policies will be ignored.

apiVersion - Must be set to consul.hashicorp.com/v1alpha1
kind - Must be set to ServiceResolver
metadata
- name - Set to the name of the service being configured.
- namespace - If running Consul Open Source, the namespace is ignored (see Kubernetes Namespaces in Consul OSS). If running Consul Enterprise see Kubernetes Namespaces in Consul Enterprise for more details.
spec
- connectTimeout (duration: 0s) - The timeout for establishing new network connections to this service. The default unit of time is ns.
- requestTimeout (duration: 0s) - The timeout for receiving an HTTP response from this service before the connection is terminated. If unspecified or 0, the default of 15s is used. The default unit of time is ns.
- defaultSubset (string: "") - The subset to use when no explicit subset is requested. If empty, the unnamed subset is used.
- subsets (map[string]ServiceResolverSubset) - A map of subset name to subset definition for all usable named subsets of this service. The map key is the name of the subset and all names must be valid DNS subdomain elements.
  
  This may be empty, in which case only the unnamed default subset will be usable.
  - filter (string: "") - The filter expression to be used for selecting instances of the requested service. If empty, all healthy instances are returned. This expression can filter on the same selectors as the Health API endpoint.
  - onlyPassing (bool: false) - Specifies the behavior of the resolver's health check interpretation. If this is set to false, instances with checks in the passing as well as the warning states will be considered healthy. If this is set to true, only instances with checks in the passing state will be considered healthy.
- redirect (ServiceResolverRedirect: <optional>) - When configured, all attempts to resolve the service this resolver defines will be substituted for the supplied redirect EXCEPT when the redirect has already been applied.
  
  When redirect is set, all other fields besides redirect will be ignored.
  - service (string: "") - A service to resolve instead of the current service.
  - serviceSubset (string: "") - A named subset of the given service to resolve instead of one defined as that service's defaultSubset. If empty, the default subset is used.
    
    If this is specified at least one of service, datacenter, or namespace should be configured.
  - namespace (string: "")Enterprise - Specifies the destination namespace to resolve the service from.
  - partition (string: "")Enterprise - Specifies the destination admin partition to resolve the service from.
  - datacenter (string: "") - Specifies the destination datacenter to resolve the service from.
  - peer (string: "") - Specifies the destination cluster peer to resolve the target service name from. Ensure that intentions are defined on the peered cluster to allow the source service to access this redirect target service as an upstream. When the peer name is specified, Consul uses Envoy's outlier detection to determine the health of the redirect target based on whether communication attempts to the redirect target are generally successful. Service health checks imported from a peer are not considered in the health of a redirect target because they do not account for service routers, splitters, and resolvers that may be defined in the peer for the target service.
- failover (map[string]ServiceResolverFailover) - Controls when and how to reroute traffic to an alternate pool of service instances.
  
  The map is keyed by the service subset it applies to and the special string "*" is a wildcard that applies to any subset not otherwise specified here.
  
  service, serviceSubset, namespace, targets and datacenters cannot all be empty at once.
  - service (string: "") - The service to resolve instead of the default as the failover group of instances during failover.
  - serviceSubset (string: "") - The named subset of the requested service to resolve as the failover group of instances. If empty, the default subset for the requested service is used.
  - namespace (string: "")Enterprise - The namespace to resolve the requested service from to form the failover group of instances. If empty, the current namespace is used.
  - datacenters (array<string>) - A fixed list of datacenters to try during failover.
  - targets (array<ServiceResolverFailoverTarget>) - A fixed list of failover targets to try during failover. It allows operators to express complicated failover scenarios such as between cluster peers, WAN federated datacenters, and local admin partitions.
    - service (string: "") - The service name to use for the failover target. If empty, the current service name is used.
    - serviceSubset (string: "") - The named subset to use for the failover target. If empty, the default subset for the requested service name is used.
    - namespace (string: "")Enterprise - The namespace to use for the failover target. If empty, the default namespace is used.
    - partition (string: "")Enterprise - The admin partition within the same datacenter to use for the failover target. If empty, the default partition is used. To use an admin partition in a different datacenter for the failover target, use the Peer field instead.
    - datacenter (string: "") - The WAN federated datacenter to use for the failover target. If empty, the current datacenter is used. To use a datacenter for the failover target that is connected with a cluster peering relationship rather than WAN federation, use the Peer field instead.
    - peer (string: "") - Specifies the destination cluster peer to resolve the target service name from. Ensure that intentions are defined on the peered cluster to allow the source service to access this failover target service as an upstream. When the peer name is specified, Consul uses Envoy's outlier detection to determine the health of the failover target based on whether communication attempts to the failover target are generally successful. Service health checks imported from a peer are not considered in the health of a failover target because they do not account for service routers, splitters, and resolvers that may be defined in the peer for the target service.
- loadBalancer (LoadBalancer) - Determines the load balancing policy and configuration for services issuing requests to this upstream. This option is available in Consul versions 1.9.0 and newer.
  - policy (string: "") - The load balancing policy used to select a host. One of: random, round_robin, least_request, ring_hash, maglev.
  - ringHashConfig (RingHashConfig) - Configuration for the ring_hash policy type.
    - minimumRingSize (int: 1024) - Determines the minimum number of entries in the hash ring.
    - maximumRingSize (int: 8192) - Determines the maximum number of entries in the hash ring.
  - leastRequestConfig (LeastRequestConfig) - Configuration for the least_request policy type.
    - choiceCount (int: 2) - Determines the number of random healthy hosts from which to select the one with the least requests.
  - hashPolicies (array<HashPolicies>) - List of hash policies to use for hashing load balancing algorithms. Hash policies are evaluated individually and combined such that identical lists result in the same hash. If no hash policies are present, or none are successfully evaluated, then a random backend host will be selected.
    - field (string: "") - The attribute type to hash on. Must be one of header, cookie, or query_parameter. Cannot be specified along with sourceIP.
    - fieldValue (string: "") - The value to hash. ie. header name, cookie name, URL query parameter name. Cannot be specified along with sourceIP.
    - cookieConfig (CookieConfig) - Additional configuration for the "cookie" hash policy type. This is specified to have Envoy generate a cookie for a client on its first request.
      - session (bool: false) - Generates a session cookie with no expiration.
      - ttl (duration: 0s) - TTL for generated cookies. Cannot be specified for session cookies.
      - path (string: "") - The path to set for the cookie.
    - sourceIP (bool: false) - Determines whether the hash should be of the source IP address rather than of a field and field value. Cannot be specified along with field or fieldValue.
    - terminal (bool: false) - Will short circuit the computation of the hash when multiple hash policies are present. If a hash is computed when a Terminal policy is evaluated, then that hash will be used and subsequent hash policies will be ignored.

Service Subsets

A service subset assigns a concrete name to a specific subset of discoverable service instances within a datacenter, such as "version2" or "canary".

A service subset name is useful only when composed with an actual service name, a specific datacenter, and namespace.

All services have an unnamed default subset that will return all healthy instances unfiltered.

Subsets are defined in service-resolver configuration entries, but are referenced by their names throughout the other configuration entry kinds.

ACLs

Configuration entries may be protected by ACLs.

Reading a service-resolver config entry requires service:read on the resource.

Creating, updating, or deleting a service-resolver config entry requires service:write on the resource and service:read on any other service referenced by name in these fields: