Add desired host phase 2 contents to OmicronSledConfig (PR 1/4) (#8538)

`HostPhase2DesiredContents` is analogous to `OmicronZoneImageSource`,
but for OS images: either keep the current contents of the boot disk or
set it to a specific artifact from the TUF repo depot. "Keep the current
contents" should show up in three cases, just like
`OmicronZoneImageSource::InstallDataset`:

1. It's the default value for deserializing, so we can load old configs
that didn't have this value
2. RSS uses it (no TUF repo depot involved at this point)
3. The planner will use this variant as a part of removing a mupdate
override (this work is still in PR itself:
#8456 (comment))

Author: jgallagher

Cargo.lock

@@ -131,6 +131,7 @@ use nexus_sled_agent_shared::inventory::BootPartitionDetails;
 use nexus_sled_agent_shared::inventory::ConfigReconcilerInventory;
 use nexus_sled_agent_shared::inventory::ConfigReconcilerInventoryResult;
 use nexus_sled_agent_shared::inventory::ConfigReconcilerInventoryStatus;
+use nexus_sled_agent_shared::inventory::HostPhase2DesiredContents;
 use nexus_sled_agent_shared::inventory::OmicronSledConfig;
 use nexus_sled_agent_shared::inventory::OmicronZoneImageSource;
 use nexus_sled_agent_shared::inventory::OrphanedDataset;
@@ -7567,12 +7568,30 @@ fn inv_collection_print_sled_config(label: &str, config: &OmicronSledConfig) {
         datasets,
         zones,
         remove_mupdate_override,
+        host_phase_2,
     } = config;
 
     println!("\n{label} SLED CONFIG");
     println!("    generation: {}", generation);
     println!("    remove_mupdate_override: {remove_mupdate_override:?}");
 
+    let display_host_phase_2_desired = |desired| match desired {
+        HostPhase2DesiredContents::CurrentContents => {
+            Cow::Borrowed("keep current contents")
+        }
+        HostPhase2DesiredContents::Artifact { hash } => {
+            Cow::Owned(format!("artifact {hash}"))
+        }
+    };
+    println!(
+        "    desired host phase 2 slot a: {}",
+        display_host_phase_2_desired(host_phase_2.slot_a)
+    );
+    println!(
+        "    desired host phase 2 slot b: {}",
+        display_host_phase_2_desired(host_phase_2.slot_b)
+    );
+
     if disks.is_empty() {
         println!("    disk config empty");
     } else {

dev-tools/omdb/src/bin/omdb/db.rs

@@ -576,6 +576,47 @@ pub enum SledRole {
     Scrimlet,
 }
 
+/// Describes the desired contents of a host phase 2 slot (i.e., the boot
+/// partition on one of the internal M.2 drives).
+#[derive(
+    Clone, Copy, Debug, Deserialize, Serialize, JsonSchema, PartialEq, Eq,
+)]
+#[serde(tag = "type", rename_all = "snake_case")]
+pub enum HostPhase2DesiredContents {
+    /// Do not change the current contents.
+    ///
+    /// We use this value when we've detected a sled has been mupdated (and we
+    /// don't want to overwrite phase 2 images until we understand how to
+    /// recover from that mupdate) and as the default value when reading an
+    /// [`OmicronSledConfig`] that was ledgered before this concept existed.
+    CurrentContents,
+
+    /// Set the phase 2 slot to the given artifact.
+    ///
+    /// The artifact will come from an unpacked and distributed TUF repo.
+    Artifact { hash: ArtifactHash },
+}
+
+/// Describes the desired contents for both host phase 2 slots.
+#[derive(Clone, Debug, Deserialize, Serialize, JsonSchema, PartialEq, Eq)]
+#[serde(rename_all = "snake_case")]
+pub struct HostPhase2DesiredSlots {
+    pub slot_a: HostPhase2DesiredContents,
+    pub slot_b: HostPhase2DesiredContents,
+}
+
+impl HostPhase2DesiredSlots {
+    /// Return a `HostPhase2DesiredSlots` with both slots set to
+    /// [`HostPhase2DesiredContents::CurrentContents`]; i.e., "make no changes
+    /// to the current contents of either slot".
+    pub const fn current_contents() -> Self {
+        Self {
+            slot_a: HostPhase2DesiredContents::CurrentContents,
+            slot_b: HostPhase2DesiredContents::CurrentContents,
+        }
+    }
+}
+
 /// Describes the set of Reconfigurator-managed configuration elements of a sled
 #[derive(Clone, Debug, Deserialize, Serialize, JsonSchema, PartialEq, Eq)]
 pub struct OmicronSledConfig {
@@ -584,6 +625,8 @@ pub struct OmicronSledConfig {
     pub datasets: IdMap<DatasetConfig>,
     pub zones: IdMap<OmicronZoneConfig>,
     pub remove_mupdate_override: Option<MupdateOverrideUuid>,
+    #[serde(default = "HostPhase2DesiredSlots::current_contents")]
+    pub host_phase_2: HostPhase2DesiredSlots,
 }
 
 impl Default for OmicronSledConfig {
@@ -594,6 +637,7 @@ impl Default for OmicronSledConfig {
             datasets: IdMap::default(),
             zones: IdMap::default(),
             remove_mupdate_override: None,
+            host_phase_2: HostPhase2DesiredSlots::current_contents(),
         }
     }
 }

nexus-sled-agent-shared/src/inventory.rs

@@ -44,6 +44,8 @@ use nexus_db_schema::schema::{
 use nexus_sled_agent_shared::inventory::BootImageHeader;
 use nexus_sled_agent_shared::inventory::BootPartitionDetails;
 use nexus_sled_agent_shared::inventory::ConfigReconcilerInventoryStatus;
+use nexus_sled_agent_shared::inventory::HostPhase2DesiredContents;
+use nexus_sled_agent_shared::inventory::HostPhase2DesiredSlots;
 use nexus_sled_agent_shared::inventory::MupdateOverrideBootInventory;
 use nexus_sled_agent_shared::inventory::MupdateOverrideInventory;
 use nexus_sled_agent_shared::inventory::MupdateOverrideNonBootInventory;
@@ -1944,6 +1946,9 @@ pub struct InvOmicronSledConfig {
     pub id: DbTypedUuid<OmicronSledConfigKind>,
     pub generation: Generation,
     pub remove_mupdate_override: Option<DbTypedUuid<MupdateOverrideKind>>,
+
+    #[diesel(embed)]
+    pub host_phase_2: DbHostPhase2DesiredSlots,
 }
 
 impl InvOmicronSledConfig {
@@ -1952,12 +1957,51 @@ impl InvOmicronSledConfig {
         id: OmicronSledConfigUuid,
         generation: external::Generation,
         remove_mupdate_override: Option<MupdateOverrideUuid>,
+        host_phase_2: HostPhase2DesiredSlots,
     ) -> Self {
         Self {
             inv_collection_id: inv_collection_id.into(),
             id: id.into(),
             generation: Generation(generation),
             remove_mupdate_override: remove_mupdate_override.map(From::from),
+            host_phase_2: host_phase_2.into(),
+        }
+    }
+}
+
+#[derive(Queryable, Clone, Debug, Selectable, Insertable)]
+#[diesel(table_name = inv_omicron_sled_config)]
+pub struct DbHostPhase2DesiredSlots {
+    pub host_phase_2_desired_slot_a: Option<ArtifactHash>,
+    pub host_phase_2_desired_slot_b: Option<ArtifactHash>,
+}
+
+impl From<HostPhase2DesiredSlots> for DbHostPhase2DesiredSlots {
+    fn from(value: HostPhase2DesiredSlots) -> Self {
+        let remap = |desired| match desired {
+            HostPhase2DesiredContents::CurrentContents => None,
+            HostPhase2DesiredContents::Artifact { hash } => {
+                Some(ArtifactHash(hash))
+            }
+        };
+        Self {
+            host_phase_2_desired_slot_a: remap(value.slot_a),
+            host_phase_2_desired_slot_b: remap(value.slot_b),
+        }
+    }
+}
+
+impl From<DbHostPhase2DesiredSlots> for HostPhase2DesiredSlots {
+    fn from(value: DbHostPhase2DesiredSlots) -> Self {
+        let remap = |maybe_artifact| match maybe_artifact {
+            None => HostPhase2DesiredContents::CurrentContents,
+            Some(ArtifactHash(hash)) => {
+                HostPhase2DesiredContents::Artifact { hash }
+            }
+        };
+        Self {
+            slot_a: remap(value.host_phase_2_desired_slot_a),
+            slot_b: remap(value.host_phase_2_desired_slot_b),
         }
     }
 }

nexus/db-model/src/inventory.rs

@@ -16,7 +16,7 @@ use std::{collections::BTreeMap, sync::LazyLock};
 ///
 /// This must be updated when you change the database schema.  Refer to
 /// schema/crdb/README.adoc in the root of this repository for details.
-pub const SCHEMA_VERSION: Version = Version::new(158, 0, 0);
+pub const SCHEMA_VERSION: Version = Version::new(159, 0, 0);
 
 /// List of all past database schema versions, in *reverse* order
 ///
@@ -28,6 +28,7 @@ static KNOWN_VERSIONS: LazyLock<Vec<KnownVersion>> = LazyLock::new(|| {
         // |  leaving the first copy as an example for the next person.
         // v
         // KnownVersion::new(next_int, "unique-dirname-with-the-sql-files"),
+        KnownVersion::new(159, "sled-config-desired-host-phase-2"),
         KnownVersion::new(158, "drop-builtin-roles"),
         KnownVersion::new(157, "user-data-export"),
         KnownVersion::new(156, "boot-partitions-inventory"),

nexus/db-model/src/schema_versions.rs

@@ -2774,6 +2774,7 @@ impl DataStore {
                             disks: IdMap::default(),
                             datasets: IdMap::default(),
                             zones: IdMap::default(),
+                            host_phase_2: sled_config.host_phase_2.into(),
                         },
                     });
                 }
@@ -3871,6 +3872,7 @@ impl ConfigReconcilerRows {
             sled_config_id,
             config.generation,
             config.remove_mupdate_override,
+            config.host_phase_2.clone(),
         ));
         self.disks.extend(config.disks.iter().map(|disk| {
             InvOmicronSledConfigDisk::new(

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

@@ -1780,6 +1780,8 @@ table! {
 
         generation -> Int8,
         remove_mupdate_override -> Nullable<Uuid>,
+        host_phase_2_desired_slot_a -> Nullable<Text>,
+        host_phase_2_desired_slot_b -> Nullable<Text>,
     }
 }
 

nexus/db-schema/src/schema.rs

@@ -32,6 +32,7 @@ sled-agent-types.workspace = true
 sled-agent-zone-images-examples.workspace = true
 slog.workspace = true
 strum.workspace = true
+swrite.workspace = true
 thiserror.workspace = true
 tufaceous-artifact.workspace = true
 typed-rng.workspace = true