Reconfigurator: Add support for boundary NTP zones (#6259)

This is almost entirely planner work; no executor work was required,
because it already supports "zones with external networking", and
there's nothing special to boundary NTP beyond that from an execution
point of view.

I put this through a fairly thorough test on london; I'll put my notes
from that in comments on the PR momentarily.

This builds on #6050 and should be merged after it.

Author: jgallagher

common/src/address.rs

@@ -25,6 +25,9 @@ pub const MAX_PORT: u16 = u16::MAX;
 /// minimum possible value for a tcp or udp port
 pub const MIN_PORT: u16 = u16::MIN;
 
+/// The amount of redundancy for boundary NTP servers.
+pub const BOUNDARY_NTP_REDUNDANCY: usize = 2;
+
 /// The amount of redundancy for Nexus services.
 ///
 /// This is used by both RSS (to distribute the initial set of services) and the

dev-tools/reconfigurator-cli/src/main.rs

@@ -34,6 +34,7 @@ use omicron_common::api::external::Generation;
 use omicron_common::api::external::Name;
 use omicron_uuid_kinds::CollectionUuid;
 use omicron_uuid_kinds::GenericUuid;
+use omicron_uuid_kinds::OmicronZoneUuid;
 use omicron_uuid_kinds::SledUuid;
 use omicron_uuid_kinds::VnicUuid;
 use reedline::{Reedline, Signal};
@@ -435,6 +436,8 @@ enum BlueprintEditCommands {
     },
     /// add a CockroachDB instance to a particular sled
     AddCockroach { sled_id: SledUuid },
+    /// expunge a particular zone from a particular sled
+    ExpungeZone { sled_id: SledUuid, zone_id: OmicronZoneUuid },
 }
 
 #[derive(Debug, Args)]
@@ -747,8 +750,8 @@ fn cmd_blueprint_edit(
 
     let label = match args.edit_command {
         BlueprintEditCommands::AddNexus { sled_id } => {
-            let current =
-                builder.sled_num_zones_of_kind(sled_id, ZoneKind::Nexus);
+            let current = builder
+                .sled_num_running_zones_of_kind(sled_id, ZoneKind::Nexus);
             let added = builder
                 .sled_ensure_zone_multiple_nexus(sled_id, current + 1)
                 .context("failed to add Nexus zone")?;
@@ -759,8 +762,8 @@ fn cmd_blueprint_edit(
             format!("added Nexus zone to sled {}", sled_id)
         }
         BlueprintEditCommands::AddCockroach { sled_id } => {
-            let current =
-                builder.sled_num_zones_of_kind(sled_id, ZoneKind::CockroachDb);
+            let current = builder
+                .sled_num_running_zones_of_kind(sled_id, ZoneKind::CockroachDb);
             let added = builder
                 .sled_ensure_zone_multiple_cockroachdb(sled_id, current + 1)
                 .context("failed to add CockroachDB zone")?;
@@ -770,9 +773,25 @@ fn cmd_blueprint_edit(
             );
             format!("added CockroachDB zone to sled {}", sled_id)
         }
+        BlueprintEditCommands::ExpungeZone { sled_id, zone_id } => {
+            builder
+                .sled_expunge_zone(sled_id, zone_id)
+                .context("failed to expunge zone")?;
+            format!("expunged zone {zone_id} from sled {sled_id}")
+        }
     };
 
-    let new_blueprint = builder.build();
+    let mut new_blueprint = builder.build();
+
+    // Normally `builder.build()` would construct the cockroach fingerprint
+    // based on what we read from CRDB and put into the planning input, but
+    // since we don't have a CRDB we had to make something up for our planning
+    // input's CRDB fingerprint. In the absense of a better alternative, we'll
+    // just copy our parent's CRDB fingerprint and carry it forward.
+    new_blueprint
+        .cockroachdb_fingerprint
+        .clone_from(&blueprint.cockroachdb_fingerprint);
+
     let rv = format!(
         "blueprint {} created from blueprint {}: {}",
         new_blueprint.id, blueprint_id, label

nexus/reconfigurator/execution/src/dns.rs

@@ -499,6 +499,7 @@ mod test {
     use omicron_common::address::get_switch_zone_address;
     use omicron_common::address::IpRange;
     use omicron_common::address::Ipv6Subnet;
+    use omicron_common::address::BOUNDARY_NTP_REDUNDANCY;
     use omicron_common::address::COCKROACHDB_REDUNDANCY;
     use omicron_common::address::NEXUS_REDUNDANCY;
     use omicron_common::address::RACK_PREFIX;
@@ -1313,6 +1314,7 @@ mod test {
                 cockroachdb_settings: &CockroachDbSettings::empty(),
                 external_ip_rows: &[],
                 service_nic_rows: &[],
+                target_boundary_ntp_zone_count: BOUNDARY_NTP_REDUNDANCY,
                 target_nexus_zone_count: NEXUS_REDUNDANCY,
                 target_cockroachdb_zone_count: COCKROACHDB_REDUNDANCY,
                 target_cockroachdb_cluster_version:
@@ -1340,7 +1342,8 @@ mod test {
         .unwrap();
         let sled_id =
             blueprint.sleds().next().expect("expected at least one sled");
-        let nalready = builder.sled_num_zones_of_kind(sled_id, ZoneKind::Nexus);
+        let nalready =
+            builder.sled_num_running_zones_of_kind(sled_id, ZoneKind::Nexus);
         let rv = builder
             .sled_ensure_zone_multiple_nexus(sled_id, nalready + 1)
             .unwrap();

nexus/reconfigurator/planning/proptest-regressions/planner/omicron_zone_placement.txt

@@ -5,3 +5,4 @@
 # It is recommended to check this file in to source control so that
 # everyone who runs the test benefits from these saved cases.
 cc 72b902d1405681df2dd46efc097da6840ff1234dc9d0d7c0ecf07bed0b0e7d8d # shrinks to input = _TestPlaceOmicronZonesArgs { input: ArbitraryTestInput { existing_sleds: {[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]: ExistingSled { zones: ZonesToPlace { zones: [] }, waiting_for_ntp: false, num_disks: 1 }}, zones_to_place: ZonesToPlace { zones: [Nexus] } } }
+cc d725ad7fd51d0409c2f24088730159c1c3043a7675d46b966e45cb86b570a141 # shrinks to input = _TestPlaceOmicronZonesArgs { input: ArbitraryTestInput { existing_sleds: {[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]: ExistingSled { zones: ZonesToPlace { zones: [] }, num_zpools: 2 }}, zones_to_place: ZonesToPlace { zones: [BoundaryNtp, BoundaryNtp] } } }