default_policy: Builder: forbid preferring racks without DCs

This is an API restriction continuing of what has already been done
in bb09756: as the case of providing
preferred rack without specifying the preferred datacenter is invalid,
in spirit of the principle “Make illegal states unrepresentable”,
`DefaultBuilder`'s API is changed accordingly to disallow creating
such configuration.
A stray backslash is deleted from a docstring.

Docs are updated to reflect the API change.

Author: wprzytula

docs/source/load-balancing/default-policy.md

@@ -10,8 +10,7 @@ for queries with non-local consistency mode is also supported.
 `builder()` method of `DefaultPolicy` returns a new instance of
 `DefaultPolicyBuilder` with the following default values:
 
-- `preferred_datacenter`: `None`
-- `preferred_rack`: `None`
+- `preferences`: no particular datacenter/rack preference
 - `is_token_aware`: `true`
 - `permit_dc_failover`: `false`
 - `latency_awareness`: `None`
@@ -25,8 +24,7 @@ You can use the builder methods to configure the desired settings and create a
 use scylla::load_balancing::DefaultPolicy;
 
 let default_policy = DefaultPolicy::builder()
-        .prefer_datacenter("dc1".to_string())
-        .prefer_rack("rack1".to_string())
+        .prefer_datacenter_and_rack("dc1".to_string(), "rack1".to_string())
         .token_aware(true)
         .permit_dc_failover(true)
         .build();
@@ -35,36 +33,39 @@ let default_policy = DefaultPolicy::builder()
 
 ### Semantics of `DefaultPolicy`
 
-#### Preferred Datacenter
+#### Preferences
+
+The `preferences` field in `DefaultPolicy` allows the load balancing
+policy to prioritize nodes based on their location. It has three modes:
+
+- no preference
+- preferred datacenter
+- preferred datacenter and rack
 
-The `preferred_datacenter` field in `DefaultPolicy` allows the load balancing
-policy to prioritize nodes based on their location. When a preferred datacenter
-is set, the policy will treat nodes in that datacenter as "local" nodes, and
-nodes in other datacenters as "remote" nodes. This affects the order in which
-nodes are returned by the policy when selecting replicas for read or write
-operations. If no preferred datacenter is specified, the policy will treat all
-nodes as local nodes.
+When a datacenter `"my_dc"` is preferred, the policy will treat nodes in `"my_dc"`
+as "local" nodes, and nodes in other datacenters as "remote" nodes. This affects
+the order in which nodes are returned by the policy when selecting replicas for
+read or write operations. If no datacenter is preferred, the policy will treat
+all nodes as local nodes.
+
+`preferences` allow the load balancing policy to prioritize nodes based on their
+availability zones (racks) in the preferred datacenter, too. When a datacenter
+and a rack are preferred, the policy will first return replicas in the local rack
+in the preferred datacenter, and then the other replicas in the datacenter
+(followed by remote replicas).
 
 When datacenter failover is disabled (`permit_dc_failover` is set to
 false), the default policy will only include local nodes in load balancing
 plans. Remote nodes will be excluded, even if they are alive and available to
 serve requests.
 
-#### Preferred Rack
-
-The `preferred_rack` field in `DefaultPolicy` allows the load balancing policy to
-prioritize nodes based on their availability zones in the preferred datacenter.
-When a preferred rack is set, the policy will first return replicas in the local rack
-in the preferred datacenter, and then the other replicas in the datacenter.
-When a preferred datacenter is not set, setting preferred rack will not have any effect.
-
 #### Datacenter Failover
 
 In the event of a datacenter outage or network failure, the nodes in that
 datacenter may become unavailable, and clients may no longer be able to access
-the data stored on those nodes. To address this, the `DefaultPolicy` supports datacenter
-failover, which allows to route requests to nodes in other datacenters if the
-local nodes are unavailable.
+the data stored on those nodes. To address this, the `DefaultPolicy` supports
+datacenter failover, which allows to route requests to nodes in other datacenters
+if the local nodes are unavailable.
 
 Datacenter failover can be enabled in `DefaultPolicy` by `permit_dc_failover`
 setting in the builder. When this flag is set, the policy will prefer to return

scylla/src/transport/load_balancing/default.rs

@@ -636,8 +636,7 @@ impl Default for DefaultPolicy {
 /// # }
 #[derive(Clone, Debug)]
 pub struct DefaultPolicyBuilder {
-    preferred_datacenter: Option<String>,
-    preferred_rack: Option<String>,
+    preferences: ReplicaLocationPreference,
     is_token_aware: bool,
     permit_dc_failover: bool,
     latency_awareness: Option<LatencyAwarenessBuilder>,
@@ -648,8 +647,7 @@ impl DefaultPolicyBuilder {
     /// Creates a builder used to customise configuration of a new DefaultPolicy.
     pub fn new() -> Self {
         Self {
-            preferred_datacenter: None,
-            preferred_rack: None,
+            preferences: ReplicaLocationPreference::Any,
             is_token_aware: true,
             permit_dc_failover: false,
             latency_awareness: None,
@@ -668,24 +666,8 @@ impl DefaultPolicyBuilder {
             Box::new(DefaultPolicy::is_alive)
         };
 
-        // As the case of providing preferred rack without providing datacenter is invalid, the rack is then ignored.
-        // According to the principle “Make illegal states unrepresentable”, in the next major release we will
-        // alter the `DefaultPolicyBuilder`'s API so that it is impossible for the user to create such state.
-        let preferences = match (self.preferred_datacenter, self.preferred_rack) {
-            (None, None) => ReplicaLocationPreference::Any,
-            (None, Some(_)) => {
-                // This a is case that the user shouldn't be able to represent.
-                warn!("Preferred rack has effect only if a preferred datacenter is set as well. Ignoring the preferred rack.");
-                ReplicaLocationPreference::Any
-            }
-            (Some(datacenter), None) => ReplicaLocationPreference::Datacenter(datacenter),
-            (Some(datacenter), Some(rack)) => {
-                ReplicaLocationPreference::DatacenterAndRack(datacenter, rack)
-            }
-        };
-
         Arc::new(DefaultPolicy {
-            preferences,
+            preferences: self.preferences,
             is_token_aware: self.is_token_aware,
             permit_dc_failover: self.permit_dc_failover,
             pick_predicate,
@@ -694,22 +676,28 @@ impl DefaultPolicyBuilder {
         })
     }
 
-    /// Sets the rack to be preferred by this policy
+    /// Sets the datacenter to be preferred by this policy.
     ///
-    /// Allows the load balancing policy to prioritize nodes based on their availability zones
-    /// in the preferred datacenter.
-    /// When a preferred rack is set, the policy will first return replicas in the local rack
-    /// in the preferred datacenter, and then the other replicas in the datacenter.
+    /// Allows the load balancing policy to prioritize nodes based on their location.
+    /// When a preferred datacenter is set, the policy will treat nodes in that
+    /// datacenter as "local" nodes, and nodes in other datacenters as "remote" nodes.
+    /// This affects the order in which nodes are returned by the policy when
+    /// selecting replicas for read or write operations. If no preferred datacenter
+    /// is specified, the policy will treat all nodes as local nodes.
     ///
-    /// When a preferred datacenter is not set, setting preferred rack will not have any effect.
-    pub fn prefer_rack(mut self, rack_name: String) -> Self {
-        self.preferred_rack = Some(rack_name);
+    /// When datacenter failover is disabled (`permit_dc_failover` is set to false),
+    /// the default policy will only include local nodes in load balancing plans.
+    /// Remote nodes will be excluded, even if they are alive and available
+    /// to serve requests.
+    pub fn prefer_datacenter(mut self, datacenter_name: String) -> Self {
+        self.preferences = ReplicaLocationPreference::Datacenter(datacenter_name);
         self
     }
 
-    /// Sets the datacenter to be preferred by this policy.
+    /// Sets the datacenter and rack to be preferred by this policy.
     ///
-    /// Allows the load balancing policy to prioritize nodes based on their location.
+    /// Allows the load balancing policy to prioritize nodes based on their location
+    /// as well as their availability zones in the preferred datacenter.
     /// When a preferred datacenter is set, the policy will treat nodes in that
     /// datacenter as "local" nodes, and nodes in other datacenters as "remote" nodes.
     /// This affects the order in which nodes are returned by the policy when
@@ -720,8 +708,15 @@ impl DefaultPolicyBuilder {
     /// the default policy will only include local nodes in load balancing plans.
     /// Remote nodes will be excluded, even if they are alive and available
     /// to serve requests.
-    pub fn prefer_datacenter(mut self, datacenter_name: String) -> Self {
-        self.preferred_datacenter = Some(datacenter_name);
+    ///
+    /// When a preferred rack is set, the policy will first return replicas in the local rack
+    /// in the preferred datacenter, and then the other replicas in the datacenter.
+    pub fn prefer_datacenter_and_rack(
+        mut self,
+        datacenter_name: String,
+        rack_name: String,
+    ) -> Self {
+        self.preferences = ReplicaLocationPreference::DatacenterAndRack(datacenter_name, rack_name);
         self
     }
 
@@ -740,7 +735,7 @@ impl DefaultPolicyBuilder {
     /// In the case of `DefaultPolicy`, token awareness is enabled by default,
     /// meaning that the policy will prefer to return alive local replicas
     /// if the token is available. This means that if the client is requesting data
-    /// that falls within the token range of a particular node, the policy will try\
+    /// that falls within the token range of a particular node, the policy will try
     /// to route the request to that node first, assuming it is alive and responsive.
     ///
     /// Token awareness can significantly improve the performance and scalability