consul/agent/structs/discovery_chain.go
R.B. Boyer 8e22d80e35
connect: fix failover through a mesh gateway to a remote datacenter (#6259)
Failover is pushed entirely down to the data plane by creating envoy
clusters and putting each successive destination in a different load
assignment priority band. For example this shows that normally requests
go to 1.2.3.4:8080 but when that fails they go to 6.7.8.9:8080:

- name: foo
  load_assignment:
    cluster_name: foo
    policy:
      overprovisioning_factor: 100000
    endpoints:
    - priority: 0
      lb_endpoints:
      - endpoint:
          address:
            socket_address:
              address: 1.2.3.4
              port_value: 8080
    - priority: 1
      lb_endpoints:
      - endpoint:
          address:
            socket_address:
              address: 6.7.8.9
              port_value: 8080

Mesh gateways route requests based solely on the SNI header tacked onto
the TLS layer. Envoy currently only lets you configure the outbound SNI
header at the cluster layer.

If you try to failover through a mesh gateway you ideally would
configure the SNI value per endpoint, but that's not possible in envoy
today.

This PR introduces a simpler way around the problem for now:

1. We identify any target of failover that will use mesh gateway mode local or
   remote and then further isolate any resolver node in the compiled discovery
   chain that has a failover destination set to one of those targets.

2. For each of these resolvers we will perform a small measurement of
   comparative healths of the endpoints that come back from the health API for the
   set of primary target and serial failover targets. We walk the list of targets
   in order and if any endpoint is healthy we return that target, otherwise we
   move on to the next target.

3. The CDS and EDS endpoints both perform the measurements in (2) for the
   affected resolver nodes.

4. For CDS this measurement selects which TLS SNI field to use for the cluster
   (note the cluster is always going to be named for the primary target)

5. For EDS this measurement selects which set of endpoints will populate the
   cluster. Priority tiered failover is ignored.

One of the big downsides to this approach to failover is that the failover
detection and correction is going to be controlled by consul rather than
deferring that entirely to the data plane as with the prior version. This also
means that we are bound to only failover using official health signals and
cannot make use of data plane signals like outlier detection to affect
failover.

In this specific scenario the lack of data plane signals is ok because the
effectiveness is already muted by the fact that the ultimate destination
endpoints will have their data plane signals scrambled when they pass through
the mesh gateway wrapper anyway so we're not losing much.

Another related fix is that we now use the endpoint health from the
underlying service, not the health of the gateway (regardless of
failover mode).
2019-08-05 13:30:35 -05:00

188 lines
5.6 KiB
Go

package structs
import (
"fmt"
"time"
)
// CompiledDiscoveryChain is the result from taking a set of related config
// entries for a single service's discovery chain and restructuring them into a
// form that is more usable for actual service discovery.
type CompiledDiscoveryChain struct {
ServiceName string
Namespace string // the namespace that the chain was compiled within
Datacenter string // the datacenter that the chain was compiled within
// CustomizationHash is a unique hash of any data that affects the
// compilation of the discovery chain other than config entries or the
// name/namespace/datacenter evaluation criteria.
//
// If set, this value should be used to prefix/suffix any generated load
// balancer data plane objects to avoid sharing customized and
// non-customized versions.
CustomizationHash string `json:",omitempty"`
// Protocol is the overall protocol shared by everything in the chain.
Protocol string `json:",omitempty"`
// StartNode is the first key into the Nodes map that should be followed
// when walking the discovery chain.
StartNode string `json:",omitempty"`
// Nodes contains all nodes available for traversal in the chain keyed by a
// unique name. You can walk this by starting with StartNode.
//
// NOTE: The names should be treated as opaque values and are only
// guaranteed to be consistent within a single compilation.
Nodes map[string]*DiscoveryGraphNode `json:",omitempty"`
// Targets is a list of all targets used in this chain.
Targets map[string]*DiscoveryTarget `json:",omitempty"`
}
func (c *CompiledDiscoveryChain) WillFailoverThroughMeshGateway(node *DiscoveryGraphNode) bool {
if node.Type != DiscoveryGraphNodeTypeResolver {
return false
}
failover := node.Resolver.Failover
if failover != nil && len(failover.Targets) > 0 {
for _, failTargetID := range failover.Targets {
failTarget := c.Targets[failTargetID]
switch failTarget.MeshGateway.Mode {
case MeshGatewayModeLocal, MeshGatewayModeRemote:
return true
}
}
}
return false
}
// IsDefault returns true if the compiled chain represents no routing, no
// splitting, and only the default resolution. We have to be careful here to
// avoid returning "yep this is default" when the only resolver action being
// applied is redirection to another resolver that is default, so we double
// check the resolver matches the requested resolver.
func (c *CompiledDiscoveryChain) IsDefault() bool {
if c.StartNode == "" || len(c.Nodes) == 0 {
return true
}
node := c.Nodes[c.StartNode]
if node == nil {
panic("not possible: missing node named '" + c.StartNode + "' in chain '" + c.ServiceName + "'")
}
if node.Type != DiscoveryGraphNodeTypeResolver {
return false
}
if !node.Resolver.Default {
return false
}
target := c.Targets[node.Resolver.Target]
return target.Service == c.ServiceName
}
const (
DiscoveryGraphNodeTypeRouter = "router"
DiscoveryGraphNodeTypeSplitter = "splitter"
DiscoveryGraphNodeTypeResolver = "resolver"
)
// DiscoveryGraphNode is a single node in the compiled discovery chain.
type DiscoveryGraphNode struct {
Type string
Name string // this is NOT necessarily a service
// fields for Type==router
Routes []*DiscoveryRoute `json:",omitempty"`
// fields for Type==splitter
Splits []*DiscoverySplit `json:",omitempty"`
// fields for Type==resolver
Resolver *DiscoveryResolver `json:",omitempty"`
}
func (s *DiscoveryGraphNode) IsRouter() bool {
return s.Type == DiscoveryGraphNodeTypeRouter
}
func (s *DiscoveryGraphNode) IsSplitter() bool {
return s.Type == DiscoveryGraphNodeTypeSplitter
}
func (s *DiscoveryGraphNode) IsResolver() bool {
return s.Type == DiscoveryGraphNodeTypeResolver
}
func (s *DiscoveryGraphNode) MapKey() string {
return fmt.Sprintf("%s:%s", s.Type, s.Name)
}
// compiled form of ServiceResolverConfigEntry
type DiscoveryResolver struct {
Default bool `json:",omitempty"`
ConnectTimeout time.Duration `json:",omitempty"`
Target string `json:",omitempty"`
Failover *DiscoveryFailover `json:",omitempty"`
}
// compiled form of ServiceRoute
type DiscoveryRoute struct {
Definition *ServiceRoute `json:",omitempty"`
NextNode string `json:",omitempty"`
}
// compiled form of ServiceSplit
type DiscoverySplit struct {
Weight float32 `json:",omitempty"`
NextNode string `json:",omitempty"`
}
// compiled form of ServiceResolverFailover
type DiscoveryFailover struct {
Targets []string `json:",omitempty"`
}
// DiscoveryTarget represents all of the inputs necessary to use a resolver
// config entry to execute a catalog query to generate a list of service
// instances during discovery.
type DiscoveryTarget struct {
ID string `json:",omitempty"`
Service string `json:",omitempty"`
ServiceSubset string `json:",omitempty"`
Namespace string `json:",omitempty"`
Datacenter string `json:",omitempty"`
MeshGateway MeshGatewayConfig `json:",omitempty"`
Subset ServiceResolverSubset `json:",omitempty"`
}
func NewDiscoveryTarget(service, serviceSubset, namespace, datacenter string) *DiscoveryTarget {
t := &DiscoveryTarget{
Service: service,
ServiceSubset: serviceSubset,
Namespace: namespace,
Datacenter: datacenter,
}
t.setID()
return t
}
func (t *DiscoveryTarget) setID() {
// NOTE: this format is similar to the SNI syntax for simplicity
if t.ServiceSubset == "" {
t.ID = fmt.Sprintf("%s.%s.%s", t.Service, t.Namespace, t.Datacenter)
} else {
t.ID = fmt.Sprintf("%s.%s.%s.%s", t.ServiceSubset, t.Service, t.Namespace, t.Datacenter)
}
}
func (t *DiscoveryTarget) String() string {
return t.ID
}