Initial support for migrating VM to a new propolis instance. (#447)

Building on the live migration work in propolis (oxidecomputer/propolis#69), this adds the nexus endpoint to trigger a migration of an instance.

Cargo.toml

@@ -46,13 +46,17 @@ panic = "abort"
 panic = "abort"
 
 #
-# It's common during development to use a local copy of dropshot or steno in the
-# parent directory.  If you want to use those, uncomment one of these blocks.
+# It's common during development to use a local copy of dropshot, propolis
+# or steno in the parent directory.  If you want to use those, uncomment
+# one of these blocks.
 #
 #[patch."https://github.com/oxidecomputer/dropshot"]
 #dropshot = { path = "../dropshot/dropshot" }
 #[patch."https://github.com/oxidecomputer/steno"]
 #steno = { path = "../steno" }
+#[patch."https://github.com/oxidecomputer/propolis"]
+#propolis-client = { path = "../propolis/client" }
+#propolis-server = { path = "../propolis/server" }
 
 #
 # Local client generation during development.

common/src/api/external/error.rs

@@ -153,6 +153,16 @@ impl Error {
         Error::InternalError { internal_message: internal_message.to_owned() }
     }
 
+    /**
+     * Generates an [`Error::InvalidRequest`] error with the specific message
+     *
+     * This should be used for failures due possibly to invalid client input
+     * or malformed requests.
+     */
+    pub fn invalid_request(message: &str) -> Error {
+        Error::InvalidRequest { message: message.to_owned() }
+    }
+
     /**
      * Generates an [`Error::ServiceUnavailable`] error with the specific
      * message

common/src/api/external/mod.rs

@@ -674,6 +674,10 @@ pub enum InstanceState {
     /// The instance is in the process of rebooting - it will remain
     /// in the "rebooting" state until the VM is starting once more.
     Rebooting,
+    /// The instance is in the process of migrating - it will remain
+    /// in the "migrating" state until the migration process is complete
+    /// and the destination propolis is ready to continue execution.
+    Migrating,
     Repairing,
     Failed,
     Destroyed,
@@ -702,6 +706,7 @@ impl TryFrom<&str> for InstanceState {
             "stopping" => InstanceState::Stopping,
             "stopped" => InstanceState::Stopped,
             "rebooting" => InstanceState::Rebooting,
+            "migrating" => InstanceState::Migrating,
             "repairing" => InstanceState::Repairing,
             "failed" => InstanceState::Failed,
             "destroyed" => InstanceState::Destroyed,
@@ -720,6 +725,7 @@ impl InstanceState {
             InstanceState::Stopping => "stopping",
             InstanceState::Stopped => "stopped",
             InstanceState::Rebooting => "rebooting",
+            InstanceState::Migrating => "migrating",
             InstanceState::Repairing => "repairing",
             InstanceState::Failed => "failed",
             InstanceState::Destroyed => "destroyed",
@@ -737,6 +743,7 @@ impl InstanceState {
             InstanceState::Running => false,
             InstanceState::Stopping => false,
             InstanceState::Rebooting => false,
+            InstanceState::Migrating => false,
 
             InstanceState::Creating => true,
             InstanceState::Stopped => true,

common/src/api/internal/nexus.rs

@@ -38,6 +38,10 @@ pub struct InstanceRuntimeState {
     pub sled_uuid: Uuid,
     /// which propolis-server is running this Instance
     pub propolis_uuid: Uuid,
+    /// address of propolis-server running this Instance
+    pub propolis_addr: Option<SocketAddr>,
+    /// migration id (if one in process)
+    pub migration_uuid: Option<Uuid>,
     /// number of CPUs allocated for this Instance
     pub ncpus: InstanceCpuCount,
     /// memory allocated for this Instance

common/src/api/internal/sled_agent.rs

@@ -8,6 +8,8 @@ use crate::api::{external, internal};
 use schemars::JsonSchema;
 use serde::{Deserialize, Serialize};
 use std::fmt::{Debug, Display, Formatter, Result as FormatResult};
+use std::net::SocketAddr;
+use uuid::Uuid;
 
 /// Describes the instance hardware.
 #[derive(Clone, Debug, Serialize, Deserialize, JsonSchema)]
@@ -24,6 +26,14 @@ pub struct InstanceEnsureBody {
     pub initial: InstanceHardware,
     /// requested runtime state of the Instance
     pub target: InstanceRuntimeStateRequested,
+    /// If we're migrating this instance, the details needed to drive the migration
+    pub migrate: Option<InstanceMigrateParams>,
+}
+
+#[derive(Clone, Debug, Deserialize, Serialize, JsonSchema)]
+pub struct InstanceMigrateParams {
+    pub src_propolis_uuid: Uuid,
+    pub src_propolis_addr: SocketAddr,
 }
 
 /// Requestable running state of an Instance.
@@ -48,6 +58,7 @@ pub enum InstanceStateRequested {
     // Issues a reset command to the instance, such that it should
     // stop and then immediately become running.
     Reboot,
+    Migrating,
     Destroyed,
 }
 
@@ -63,6 +74,7 @@ impl InstanceStateRequested {
             InstanceStateRequested::Running => "running",
             InstanceStateRequested::Stopped => "stopped",
             InstanceStateRequested::Reboot => "reboot",
+            InstanceStateRequested::Migrating => "migrating",
             InstanceStateRequested::Destroyed => "destroyed",
         }
     }
@@ -73,16 +85,19 @@ impl InstanceStateRequested {
             InstanceStateRequested::Running => false,
             InstanceStateRequested::Stopped => true,
             InstanceStateRequested::Reboot => false,
+            InstanceStateRequested::Migrating => false,
             InstanceStateRequested::Destroyed => true,
         }
     }
 }
 
 /// Used to request an Instance state change from a sled agent
 ///
-/// Right now, it's only the run state that can be changed, though we might want
-/// to support changing properties like "ncpus" here.
+/// Right now, it's only the run state and migration id that can
+/// be changed, though we might want to support changing properties
+/// like "ncpus" here.
 #[derive(Clone, Debug, Deserialize, Serialize, JsonSchema)]
 pub struct InstanceRuntimeStateRequested {
     pub run_state: InstanceStateRequested,
+    pub migration_id: Option<Uuid>,
 }

common/src/sql/dbinit.sql

@@ -253,6 +253,12 @@ CREATE TABLE omicron.public.instance (
     active_server_id UUID,
     /* Identifies the underlying propolis-server backing the instance. */
     active_propolis_id UUID,
+    active_propolis_ip INET,
+
+    /*
+     * Identifies an ongoing migration for this instance.
+     */
+    migration_id UUID,
 
     /* Instance configuration */
     ncpus INT NOT NULL,

nexus-client/src/lib.rs

@@ -50,6 +50,7 @@ impl From<types::InstanceState>
             types::InstanceState::Stopping => Self::Stopping,
             types::InstanceState::Stopped => Self::Stopped,
             types::InstanceState::Rebooting => Self::Rebooting,
+            types::InstanceState::Migrating => Self::Migrating,
             types::InstanceState::Repairing => Self::Repairing,
             types::InstanceState::Failed => Self::Failed,
             types::InstanceState::Destroyed => Self::Destroyed,
@@ -67,6 +68,8 @@ impl From<omicron_common::api::internal::nexus::InstanceRuntimeState>
             run_state: s.run_state.into(),
             sled_uuid: s.sled_uuid,
             propolis_uuid: s.propolis_uuid,
+            propolis_addr: s.propolis_addr.map(|addr| addr.to_string()),
+            migration_uuid: s.migration_uuid,
             ncpus: s.ncpus.into(),
             memory: s.memory.into(),
             hostname: s.hostname,
@@ -86,6 +89,8 @@ impl From<&omicron_common::api::internal::nexus::InstanceRuntimeState>
             run_state: s.run_state.into(),
             sled_uuid: s.sled_uuid,
             propolis_uuid: s.propolis_uuid,
+            propolis_addr: s.propolis_addr.map(|addr| addr.to_string()),
+            migration_uuid: s.migration_uuid,
             ncpus: s.ncpus.into(),
             memory: s.memory.into(),
             hostname: s.hostname.clone(),
@@ -118,6 +123,9 @@ impl From<omicron_common::api::external::InstanceState>
             omicron_common::api::external::InstanceState::Rebooting => {
                 Self::Rebooting
             }
+            omicron_common::api::external::InstanceState::Migrating => {
+                Self::Migrating
+            }
             omicron_common::api::external::InstanceState::Repairing => {
                 Self::Repairing
             }
@@ -194,6 +202,7 @@ impl From<&types::InstanceState>
             types::InstanceState::Stopping => Self::Stopping,
             types::InstanceState::Stopped => Self::Stopped,
             types::InstanceState::Rebooting => Self::Rebooting,
+            types::InstanceState::Migrating => Self::Migrating,
             types::InstanceState::Repairing => Self::Repairing,
             types::InstanceState::Failed => Self::Failed,
             types::InstanceState::Destroyed => Self::Destroyed,