[reconfigurator] Add RPW to reconcile debug dataset rendezvous table (#7342)

This is PR 2 of 2 and builds on #7341; it adds an RPW that calls the
library added in that PR to actually reconcile blueprint+inventory and
update the debug dataset rendezvous table, and changes the support
bundle query that picks a debug dataset to use this new table.

Author: jgallagher

Cargo.lock

@@ -43,6 +43,11 @@ task: "blueprint_loader"
     Loads the current target blueprint from the DB
 
 
+task: "blueprint_rendezvous"
+    reconciles blueprints and inventory collection, updating Reconfigurator-
+    owned rendezvous tables that other subsystems consume
+
+
 task: "crdb_node_id_collector"
     Collects node IDs of running CockroachDB zones
 
@@ -218,6 +223,11 @@ task: "blueprint_loader"
     Loads the current target blueprint from the DB
 
 
+task: "blueprint_rendezvous"
+    reconciles blueprints and inventory collection, updating Reconfigurator-
+    owned rendezvous tables that other subsystems consume
+
+
 task: "crdb_node_id_collector"
     Collects node IDs of running CockroachDB zones
 
@@ -380,6 +390,11 @@ task: "blueprint_loader"
     Loads the current target blueprint from the DB
 
 
+task: "blueprint_rendezvous"
+    reconciles blueprints and inventory collection, updating Reconfigurator-
+    owned rendezvous tables that other subsystems consume
+
+
 task: "crdb_node_id_collector"
     Collects node IDs of running CockroachDB zones
 

dev-tools/omdb/tests/env.out

@@ -262,6 +262,11 @@ task: "blueprint_loader"
     Loads the current target blueprint from the DB
 
 
+task: "blueprint_rendezvous"
+    reconciles blueprints and inventory collection, updating Reconfigurator-
+    owned rendezvous tables that other subsystems consume
+
+
 task: "crdb_node_id_collector"
     Collects node IDs of running CockroachDB zones
 
@@ -499,6 +504,13 @@ task: "bfd_manager"
     started at <REDACTED_TIMESTAMP> (<REDACTED DURATION>s ago) and ran for <REDACTED DURATION>ms
     last completion reported error: failed to resolve addresses for Dendrite services: no record found for Query { name: Name("_dendrite._tcp.control-plane.oxide.internal."), query_type: SRV, query_class: IN }
 
+task: "blueprint_rendezvous"
+  configured period: every <REDACTED_DURATION>m
+  currently executing: no
+  last completed activation: <REDACTED ITERATIONS>, triggered by a dependent task completing
+    started at <REDACTED_TIMESTAMP> (<REDACTED DURATION>s ago) and ran for <REDACTED DURATION>ms
+    last completion reported error: no blueprint
+
 task: "crdb_node_id_collector"
   configured period: every <REDACTED_DURATION>m
   currently executing: no
@@ -951,6 +963,13 @@ task: "bfd_manager"
     started at <REDACTED_TIMESTAMP> (<REDACTED DURATION>s ago) and ran for <REDACTED DURATION>ms
     last completion reported error: failed to resolve addresses for Dendrite services: no record found for Query { name: Name("_dendrite._tcp.control-plane.oxide.internal."), query_type: SRV, query_class: IN }
 
+task: "blueprint_rendezvous"
+  configured period: every <REDACTED_DURATION>m
+  currently executing: no
+  last completed activation: <REDACTED ITERATIONS>, triggered by a dependent task completing
+    started at <REDACTED_TIMESTAMP> (<REDACTED DURATION>s ago) and ran for <REDACTED DURATION>ms
+    last completion reported error: no blueprint
+
 task: "crdb_node_id_collector"
   configured period: every <REDACTED_DURATION>m
   currently executing: no

dev-tools/omdb/tests/successes.out

@@ -580,6 +580,12 @@ pub struct BlueprintTasksConfig {
     #[serde_as(as = "DurationSeconds<u64>")]
     pub period_secs_execute: Duration,
 
+    /// period (in seconds) for periodic activations of the background task that
+    /// reconciles the latest blueprint and latest inventory collection into
+    /// Rencofigurator rendezvous tables
+    #[serde_as(as = "DurationSeconds<u64>")]
+    pub period_secs_rendezvous: Duration,
+
     /// period (in seconds) for periodic activations of the background task that
     /// collects the node IDs of CockroachDB zones
     #[serde_as(as = "DurationSeconds<u64>")]
@@ -953,6 +959,7 @@ mod test {
             phantom_disks.period_secs = 30
             blueprints.period_secs_load = 10
             blueprints.period_secs_execute = 60
+            blueprints.period_secs_rendezvous = 300
             blueprints.period_secs_collect_crdb_node_ids = 180
             sync_service_zone_nat.period_secs = 30
             switch_port_settings_manager.period_secs = 30
@@ -1108,6 +1115,7 @@ mod test {
                             period_secs_execute: Duration::from_secs(60),
                             period_secs_collect_crdb_node_ids:
                                 Duration::from_secs(180),
+                            period_secs_rendezvous: Duration::from_secs(300),
                         },
                         sync_service_zone_nat: SyncServiceZoneNatConfig {
                             period_secs: Duration::from_secs(30)
@@ -1231,6 +1239,7 @@ mod test {
             phantom_disks.period_secs = 30
             blueprints.period_secs_load = 10
             blueprints.period_secs_execute = 60
+            blueprints.period_secs_rendezvous = 300
             blueprints.period_secs_collect_crdb_node_ids = 180
             sync_service_zone_nat.period_secs = 30
             switch_port_settings_manager.period_secs = 30

nexus-config/src/nexus_config.rs

@@ -109,6 +109,7 @@ nexus-metrics-producer-gc.workspace = true
 nexus-reconfigurator-execution.workspace = true
 nexus-reconfigurator-planning.workspace = true
 nexus-reconfigurator-preparation.workspace = true
+nexus-reconfigurator-rendezvous.workspace = true
 nexus-sled-agent-shared.workspace = true
 nexus-types.workspace = true
 omicron-common.workspace = true

nexus/Cargo.toml

@@ -2032,6 +2032,7 @@ allow_tables_to_appear_in_same_query!(hw_baseboard_id, inv_sled_agent,);
 allow_tables_to_appear_in_same_query!(
     bp_omicron_zone,
     bp_target,
+    rendezvous_debug_dataset,
     dataset,
     disk,
     image,

nexus/db-model/src/schema.rs

@@ -11,8 +11,7 @@ use crate::db;
 use crate::db::error::public_error_from_diesel;
 use crate::db::error::ErrorHandler;
 use crate::db::lookup::LookupPath;
-use crate::db::model::Dataset;
-use crate::db::model::DatasetKind;
+use crate::db::model::RendezvousDebugDataset;
 use crate::db::model::SupportBundle;
 use crate::db::model::SupportBundleState;
 use crate::db::pagination::paginated;
@@ -21,7 +20,6 @@ use crate::transaction_retry::OptionalError;
 use async_bb8_diesel::AsyncRunQueryDsl;
 use diesel::prelude::*;
 use futures::FutureExt;
-use nexus_types::identity::Asset;
 use omicron_common::api::external;
 use omicron_common::api::external::CreateResult;
 use omicron_common::api::external::DataPageParams;
@@ -31,7 +29,6 @@ use omicron_common::api::external::LookupResult;
 use omicron_uuid_kinds::GenericUuid;
 use omicron_uuid_kinds::OmicronZoneUuid;
 use omicron_uuid_kinds::SupportBundleUuid;
-use omicron_uuid_kinds::ZpoolUuid;
 use uuid::Uuid;
 
 const CANNOT_ALLOCATE_ERR_MSG: &'static str =
@@ -93,21 +90,20 @@ impl DataStore {
                 let err = err.clone();
 
                 async move {
-                    use db::schema::dataset::dsl as dataset_dsl;
+                    use db::schema::rendezvous_debug_dataset::dsl as dataset_dsl;
                     use db::schema::support_bundle::dsl as support_bundle_dsl;
 
                     // Observe all "non-deleted, debug datasets".
                     //
                     // Return the first one we find that doesn't already
                     // have a support bundle allocated to it.
-                    let free_dataset = dataset_dsl::dataset
-                        .filter(dataset_dsl::time_deleted.is_null())
-                        .filter(dataset_dsl::kind.eq(DatasetKind::Debug))
+                    let free_dataset = dataset_dsl::rendezvous_debug_dataset
+                        .filter(dataset_dsl::time_tombstoned.is_null())
                         .left_join(support_bundle_dsl::support_bundle.on(
                             dataset_dsl::id.eq(support_bundle_dsl::dataset_id),
                         ))
                         .filter(support_bundle_dsl::dataset_id.is_null())
-                        .select(Dataset::as_select())
+                        .select(RendezvousDebugDataset::as_select())
                         .first_async(&conn)
                         .await
                         .optional()?;
@@ -129,7 +125,7 @@ impl DataStore {
 
                     let bundle = SupportBundle::new(
                         reason_for_creation,
-                        ZpoolUuid::from_untyped_uuid(dataset.pool_id),
+                        dataset.pool_id(),
                         dataset.id(),
                         this_nexus_id,
                     );
@@ -499,6 +495,7 @@ mod test {
     use omicron_uuid_kinds::DatasetUuid;
     use omicron_uuid_kinds::PhysicalDiskUuid;
     use omicron_uuid_kinds::SledUuid;
+    use omicron_uuid_kinds::ZpoolUuid;
     use rand::Rng;
 
     fn authz_support_bundle_from_id(
@@ -569,6 +566,7 @@ mod test {
             opctx: &OpContext,
         ) {
             let rack_id = Uuid::new_v4();
+            let blueprint_id = BlueprintUuid::new_v4();
             let sled = SledUpdate::new(
                 *self.sled.as_untyped_uuid(),
                 "[::1]:0".parse().unwrap(),
@@ -601,18 +599,17 @@ mod test {
                 datastore
                     .zpool_insert(opctx, zpool)
                     .await
-                    .expect("failed to upsert zpool");
+                    .expect("inserted zpool");
 
-                let dataset = Dataset::new(
+                let dataset = RendezvousDebugDataset::new(
                     pool.dataset,
-                    pool.pool.into_untyped_uuid(),
-                    None,
-                    DebugDatasetKind,
+                    pool.pool,
+                    blueprint_id,
                 );
                 datastore
-                    .dataset_upsert(dataset)
+                    .debug_dataset_insert_if_not_exists(opctx, dataset)
                     .await
-                    .expect("failed to upsert dataset");
+                    .expect("inserted debug dataset");
             }
         }
     }

nexus/db-queries/src/db/datastore/support_bundle.rs

@@ -126,6 +126,7 @@ support_bundle_collector.period_secs = 30
 decommissioned_disk_cleaner.period_secs = 60
 blueprints.period_secs_load = 10
 blueprints.period_secs_execute = 60
+blueprints.period_secs_rendezvous = 300
 blueprints.period_secs_collect_crdb_node_ids = 180
 sync_service_zone_nat.period_secs = 30
 switch_port_settings_manager.period_secs = 30

nexus/examples/config-second.toml

@@ -112,6 +112,7 @@ support_bundle_collector.period_secs = 30
 decommissioned_disk_cleaner.period_secs = 60
 blueprints.period_secs_load = 10
 blueprints.period_secs_execute = 60
+blueprints.period_secs_rendezvous = 300
 blueprints.period_secs_collect_crdb_node_ids = 180
 sync_service_zone_nat.period_secs = 30
 switch_port_settings_manager.period_secs = 30

nexus/examples/config.toml

@@ -92,6 +92,7 @@ use super::tasks::abandoned_vmm_reaper;
 use super::tasks::bfd;
 use super::tasks::blueprint_execution;
 use super::tasks::blueprint_load;
+use super::tasks::blueprint_rendezvous;
 use super::tasks::crdb_node_id_collector;
 use super::tasks::decommissioned_disk_cleaner;
 use super::tasks::dns_config;
@@ -156,6 +157,7 @@ pub struct BackgroundTasks {
     pub task_phantom_disks: Activator,
     pub task_blueprint_loader: Activator,
     pub task_blueprint_executor: Activator,
+    pub task_blueprint_rendezvous: Activator,
     pub task_crdb_node_id_collector: Activator,
     pub task_service_zone_nat_tracker: Activator,
     pub task_switch_port_settings_manager: Activator,
@@ -243,6 +245,7 @@ impl BackgroundTasksInitializer {
             task_phantom_disks: Activator::new(),
             task_blueprint_loader: Activator::new(),
             task_blueprint_executor: Activator::new(),
+            task_blueprint_rendezvous: Activator::new(),
             task_crdb_node_id_collector: Activator::new(),
             task_service_zone_nat_tracker: Activator::new(),
             task_switch_port_settings_manager: Activator::new(),
@@ -311,6 +314,7 @@ impl BackgroundTasksInitializer {
             task_phantom_disks,
             task_blueprint_loader,
             task_blueprint_executor,
+            task_blueprint_rendezvous,
             task_crdb_node_id_collector,
             task_service_zone_nat_tracker,
             task_switch_port_settings_manager,
@@ -491,19 +495,15 @@ impl BackgroundTasksInitializer {
             period: config.blueprints.period_secs_collect_crdb_node_ids,
             task_impl: Box::new(crdb_node_id_collector),
             opctx: opctx.child(BTreeMap::new()),
-            watchers: vec![Box::new(rx_blueprint)],
+            watchers: vec![Box::new(rx_blueprint.clone())],
             activator: task_crdb_node_id_collector,
         });
 
         // Background task: inventory collector
         //
-        // This currently depends on the "output" of the blueprint executor in
+        // This depends on the "output" of the blueprint executor in
         // order to automatically trigger inventory collection whenever the
-        // blueprint executor runs.  In the limit, this could become a problem
-        // because the blueprint executor might also depend indirectly on the
-        // inventory collector.  In that case, we could expose `Activator`s to
-        // one or both of these tasks to directly activate the other precisely
-        // when needed.  But for now, this works.
+        // blueprint executor runs.
         let inventory_watcher = {
             let collector = inventory_collection::InventoryCollector::new(
                 datastore.clone(),
@@ -563,10 +563,28 @@ impl BackgroundTasksInitializer {
                 ),
             ),
             opctx: opctx.child(BTreeMap::new()),
-            watchers: vec![Box::new(inventory_watcher)],
+            watchers: vec![Box::new(inventory_watcher.clone())],
             activator: task_physical_disk_adoption,
         });
 
+        driver.register(TaskDefinition {
+            name: "blueprint_rendezvous",
+            description:
+                "reconciles blueprints and inventory collection, updating \
+                 Reconfigurator-owned rendezvous tables that other subsystems \
+                 consume",
+            period: config.blueprints.period_secs_rendezvous,
+            task_impl: Box::new(
+                blueprint_rendezvous::BlueprintRendezvous::new(
+                    datastore.clone(),
+                    rx_blueprint.clone(),
+                ),
+            ),
+            opctx: opctx.child(BTreeMap::new()),
+            watchers: vec![Box::new(inventory_watcher.clone())],
+            activator: task_blueprint_rendezvous,
+        });
+
         driver.register(TaskDefinition {
             name: "decommissioned_disk_cleaner",
             description: