Add sled-agent endpoint for fetching sled identifiers (#6086)

- Adds a `SledIdentifiers` type, with the most salient bits of
identifying metadata for a single sled
- Adds sled-agent endpoint `/sled-identifiers` for fetching the above
from the sled. The main goal is to provide the main data for #5267,
attaching sled identifiers to most timeseries.
- Small improvement to instructions in package-manifest.toml.
- Small bugfix to error-reporting in `xtask download`.

clients/sled-agent-client/src/lib.rs

@@ -593,6 +593,36 @@ impl From<omicron_common::api::internal::shared::NetworkInterfaceKind>
     }
 }
 
+impl From<omicron_common::api::internal::shared::SledIdentifiers>
+    for types::SledIdentifiers
+{
+    fn from(
+        value: omicron_common::api::internal::shared::SledIdentifiers,
+    ) -> Self {
+        Self {
+            model: value.model,
+            rack_id: value.rack_id,
+            revision: value.revision,
+            serial: value.serial,
+            sled_id: value.sled_id,
+        }
+    }
+}
+
+impl From<types::SledIdentifiers>
+    for omicron_common::api::internal::shared::SledIdentifiers
+{
+    fn from(value: types::SledIdentifiers) -> Self {
+        Self {
+            model: value.model,
+            rack_id: value.rack_id,
+            revision: value.revision,
+            serial: value.serial,
+            sled_id: value.sled_id,
+        }
+    }
+}
+
 /// Exposes additional [`Client`] interfaces for use by the test suite. These
 /// are bonus endpoints, not generated in the real client.
 #[async_trait]

common/src/api/internal/shared.rs

@@ -702,6 +702,26 @@ pub struct ResolvedVpcRouteSet {
     pub routes: HashSet<ResolvedVpcRoute>,
 }
 
+/// Identifiers for a single sled.
+///
+/// This is intended primarily to be used in timeseries, to identify
+/// sled from which metric data originates.
+#[derive(Clone, Debug, Deserialize, Serialize, JsonSchema, PartialEq)]
+pub struct SledIdentifiers {
+    /// Control plane ID of the rack this sled is a member of
+    pub rack_id: Uuid,
+    /// Control plane ID for the sled itself
+    pub sled_id: Uuid,
+    /// Model name of the sled
+    pub model: String,
+    /// Revision number of the sled
+    pub revision: u32,
+    /// Serial number of the sled
+    //
+    // NOTE: This is only guaranteed to be unique within a model.
+    pub serial: String,
+}
+
 #[cfg(test)]
 mod tests {
     use crate::api::internal::shared::AllowedSourceIps;

dev-tools/xtask/src/download.rs

@@ -242,7 +242,7 @@ async fn get_values_from_file<const N: usize>(
 
     let content = tokio::fs::read_to_string(&path)
         .await
-        .context("Failed to read {path}")?;
+        .with_context(|| format!("Failed to read {path}"))?;
     for line in content.lines() {
         let line = line.trim();
         let Some((key, value)) = line.split_once('=') else {

openapi/sled-agent.json

@@ -710,6 +710,30 @@
         }
       }
     },
+    "/sled-identifiers": {
+      "get": {
+        "summary": "Fetch sled identifiers",
+        "operationId": "sled_identifiers",
+        "responses": {
+          "200": {
+            "description": "successful operation",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/SledIdentifiers"
+                }
+              }
+            }
+          },
+          "4XX": {
+            "$ref": "#/components/responses/Error"
+          },
+          "5XX": {
+            "$ref": "#/components/responses/Error"
+          }
+        }
+      }
+    },
     "/sled-role": {
       "get": {
         "operationId": "sled_role_get",
@@ -4549,6 +4573,43 @@
         "type": "string",
         "pattern": "^(0|[1-9]\\d*)\\.(0|[1-9]\\d*)\\.(0|[1-9]\\d*)(?:-((?:0|[1-9]\\d*|\\d*[a-zA-Z-][0-9a-zA-Z-]*)(?:\\.(?:0|[1-9]\\d*|\\d*[a-zA-Z-][0-9a-zA-Z-]*))*))?(?:\\+([0-9a-zA-Z-]+(?:\\.[0-9a-zA-Z-]+)*))?$"
       },
+      "SledIdentifiers": {
+        "description": "Identifiers for a single sled.\n\nThis is intended primarily to be used in timeseries, to identify sled from which metric data originates.",
+        "type": "object",
+        "properties": {
+          "model": {
+            "description": "Model name of the sled",
+            "type": "string"
+          },
+          "rack_id": {
+            "description": "Control plane ID of the rack this sled is a member of",
+            "type": "string",
+            "format": "uuid"
+          },
+          "revision": {
+            "description": "Revision number of the sled",
+            "type": "integer",
+            "format": "uint32",
+            "minimum": 0
+          },
+          "serial": {
+            "description": "Serial number of the sled",
+            "type": "string"
+          },
+          "sled_id": {
+            "description": "Control plane ID for the sled itself",
+            "type": "string",
+            "format": "uuid"
+          }
+        },
+        "required": [
+          "model",
+          "rack_id",
+          "revision",
+          "serial",
+          "sled_id"
+        ]
+      },
       "SledInstanceState": {
         "description": "A wrapper type containing a sled's total knowledge of the state of a specific VMM and the instance it incarnates.",
         "type": "object",

package-manifest.toml

@@ -639,8 +639,10 @@ only_for_targets.image = "standard"
 # 1. Build the zone image manually
 #   1a. cd <dendrite tree>
 #   1b. cargo build --features=tofino_stub --release
-#   1c. cargo xtask dist -o -r --features tofino_stub
+#   1c. cargo xtask dist --format omicron --release --features tofino_stub
 # 2. Copy dendrite.tar.gz from dendrite/out to omicron/out
+# 3. Change the below `source.type` key to `"manual"` and comment out or remove
+# the other `source.*` keys.
 source.type = "prebuilt"
 source.repo = "dendrite"
 source.commit = "e83f4f164fd3dbb2100989a399a4fa087232ac36"
@@ -664,8 +666,10 @@ only_for_targets.image = "standard"
 # 1. Build the zone image manually
 #   1a. cd <dendrite tree>
 #   1b. cargo build --features=tofino_asic --release
-#   1c. cargo xtask dist -o -r --features tofino_asic
+#   1c. cargo xtask dist --format omicron --release --features tofino_asic
 # 2. Copy the output zone image from dendrite/out to omicron/out
+# 3. Change the below `source.type` key to `"manual"` and comment out or remove
+# the other `source.*` keys.
 source.type = "prebuilt"
 source.repo = "dendrite"
 source.commit = "e83f4f164fd3dbb2100989a399a4fa087232ac36"
@@ -682,8 +686,10 @@ only_for_targets.image = "standard"
 # 1. Build the zone image manually
 #   1a. cd <dendrite tree>
 #   1b. cargo build --features=softnpu --release
-#   1c. cargo xtask dist -o -r --features softnpu
+#   1c. cargo xtask dist --format omicron --release --features softnpu
 # 2. Copy dendrite.tar.gz from dendrite/out to omicron/out/dendrite-softnpu.tar.gz
+# 3. Change the below `source.type` key to `"manual"` and comment out or remove
+# the other `source.*` keys.
 source.type = "prebuilt"
 source.repo = "dendrite"
 source.commit = "e83f4f164fd3dbb2100989a399a4fa087232ac36"

sled-agent/src/http_entrypoints.rs

@@ -32,7 +32,7 @@ use omicron_common::api::internal::nexus::{
     DiskRuntimeState, SledInstanceState, UpdateArtifactId,
 };
 use omicron_common::api::internal::shared::{
-    ResolvedVpcRouteSet, ResolvedVpcRouteState, SwitchPorts,
+    ResolvedVpcRouteSet, ResolvedVpcRouteState, SledIdentifiers, SwitchPorts,
 };
 use omicron_uuid_kinds::{GenericUuid, InstanceUuid};
 use schemars::JsonSchema;
@@ -89,6 +89,7 @@ pub fn api() -> SledApiDescription {
         api.register(host_os_write_status_get)?;
         api.register(host_os_write_status_delete)?;
         api.register(inventory)?;
+        api.register(sled_identifiers)?;
         api.register(bootstore_status)?;
         api.register(list_vpc_routes)?;
         api.register(set_vpc_routes)?;
@@ -1012,6 +1013,22 @@ async fn inventory(
     Ok(HttpResponseOk(sa.inventory().await?))
 }
 
+/// Fetch sled identifiers
+#[endpoint {
+    method = GET,
+    path = "/sled-identifiers",
+}]
+async fn sled_identifiers(
+    request_context: RequestContext<SledAgent>,
+) -> Result<HttpResponseOk<SledIdentifiers>, HttpError> {
+    request_context
+        .context()
+        .sled_identifiers()
+        .await
+        .map(HttpResponseOk)
+        .map_err(HttpError::from)
+}
+
 /// Get the internal state of the local bootstore node
 #[endpoint {
     method = GET,

sled-agent/src/sled_agent.rs

@@ -51,7 +51,7 @@ use omicron_common::api::internal::nexus::{
 };
 use omicron_common::api::internal::shared::{
     HostPortConfig, RackNetworkConfig, ResolvedVpcRouteSet,
-    ResolvedVpcRouteState,
+    ResolvedVpcRouteState, SledIdentifiers,
 };
 use omicron_common::api::{
     internal::nexus::DiskRuntimeState, internal::nexus::InstanceRuntimeState,
@@ -156,6 +156,9 @@ pub enum Error {
 
     #[error("Metrics error: {0}")]
     Metrics(#[from] crate::metrics::Error),
+
+    #[error("Expected revision to fit in a u32, but found {0}")]
+    UnexpectedRevision(i64),
 }
 
 impl From<Error> for omicron_common::api::external::Error {
@@ -1185,6 +1188,31 @@ impl SledAgent {
         &self.inner.boot_disk_os_writer
     }
 
+    /// Return identifiers for this sled.
+    ///
+    /// This is mostly used to identify timeseries data with the originating
+    /// sled.
+    ///
+    /// NOTE: This only returns the identifiers for the _sled_ itself. If you're
+    /// interested in the switch identifiers, MGS is the current best way to do
+    /// that, by asking for the local switch's slot, and then that switch's SP
+    /// state.
+    pub(crate) async fn sled_identifiers(
+        &self,
+    ) -> Result<SledIdentifiers, Error> {
+        let baseboard = self.inner.hardware.baseboard();
+        Ok(SledIdentifiers {
+            rack_id: self.inner.start_request.body.rack_id,
+            sled_id: self.inner.id,
+            model: baseboard.model().to_string(),
+            revision: baseboard
+                .revision()
+                .try_into()
+                .map_err(|_| Error::UnexpectedRevision(baseboard.revision()))?,
+            serial: baseboard.identifier().to_string(),
+        })
+    }
+
     /// Return basic information about ourselves: identity and status
     ///
     /// This is basically a GET version of the information we push to Nexus on