Define timeseries schema in TOML

This is the first commit in a thread of work to support updates to timeseries schema. In our first step, we're moving the definition of timeseries from the Rust structs we use today, to TOML text files. Each file describes one target and all the metrics associated with it, from which we generate exactly the same Rust code that it replaces. This opens the door to a lot of downstream work, including well-defined updates to schema; detection of conflicting schema; improved metadata for timeseries; and generation of documentation. As an example, this moves exactly one (1) timeseries into the new format, the physical datalink statistics tracked through illumos kstats. - Move the `oximeter` crate into a private implementation crate, and re-export it at its original path - Add intermediate representation of timeseries schema, and code for deserializing TOML timeseries definitions, checking them for conflicts, and handling updates. Note that **no actual updates** are currently supported. The ingesting code includes a check that version numbers are exactly 1, except in tests. We include tests that check updates in a number of ways, but until the remainder of the work is done, we limit timeseries in the wild to their current version of 1. The actual file format also has version number associated with it, currently asserted to be 1, so that we can more easily update the entire format if needed. - Add a macro for consuming timeseries definitions in TOML, and generating the equivalent Rust code. Developers should use this new `oximeter::use_timeseries!()` proc macro to create new timeseries. - Add an integration test in `oximeter` that pulls in all the centrally-defined timeseries schema, and checks them all for conflicts. As developers add new timeseries in TOML format, they will be picked up automatically by this test, to detect problems at merge time. - Updates `kstat-rs` dep to simplify platform-specific code. This is part of this commit because we need the definitions of the physical-data-link timeseries for our integration test, but those were previously only compiled on illumos platforms. They're now included everywhere, while the tests remain illumos-only.
oxidecomputer · Jun 25, 2024 · b1be993 · b1be993
1 parent 86e1710
commit b1be993
Show file tree

Hide file tree

Showing 41 changed files with 3,249 additions and 544 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -59,10 +59,12 @@ members = [
     "nexus/types",
     "oximeter/collector",
     "oximeter/db",
+    "oximeter/impl",
     "oximeter/instruments",
     "oximeter/oximeter-macro-impl",
     "oximeter/oximeter",
     "oximeter/producer",
+    "oximeter/timeseries-macro",
     "package",
     "passwords",
     "rpaths",
@@ -149,10 +151,12 @@ default-members = [
     "nexus/types",
     "oximeter/collector",
     "oximeter/db",
+    "oximeter/impl",
     "oximeter/instruments",
     "oximeter/oximeter-macro-impl",
     "oximeter/oximeter",
     "oximeter/producer",
+    "oximeter/timeseries-macro",
     "package",
     "passwords",
     "rpaths",
@@ -382,9 +386,11 @@ oximeter = { path = "oximeter/oximeter" }
 oximeter-client = { path = "clients/oximeter-client" }
 oximeter-db = { path = "oximeter/db/" }
 oximeter-collector = { path = "oximeter/collector" }
+oximeter-impl = { path = "oximeter/impl" }
 oximeter-instruments = { path = "oximeter/instruments" }
 oximeter-macro-impl = { path = "oximeter/oximeter-macro-impl" }
 oximeter-producer = { path = "oximeter/producer" }
+oximeter-timeseries-macro = { path = "oximeter/timeseries-macro" }
 p256 = "0.13"
 parse-display = "0.9.0"
 partial-io = { version = "0.5.4", features = ["proptest1", "tokio1"] }

diff --git a/nexus/Cargo.toml b/nexus/Cargo.toml
@@ -135,7 +135,7 @@ rcgen.workspace = true
 regex.workspace = true
 similar-asserts.workspace = true
 sp-sim.workspace = true
-rustls = { workspace = true }
+rustls.workspace = true
 subprocess.workspace = true
 term.workspace = true
 trust-dns-resolver.workspace = true

diff --git a/openapi/bootstrap-agent.json b/openapi/bootstrap-agent.json
@@ -328,6 +328,7 @@
           "checker": {
             "nullable": true,
             "description": "Checker to apply to incoming messages.",
+            "default": null,
             "type": "string"
           },
           "originate": {
@@ -340,6 +341,7 @@
           "shaper": {
             "nullable": true,
             "description": "Shaper to apply to outgoing messages.",
+            "default": null,
             "type": "string"
           }
         },
@@ -437,25 +439,29 @@
           "local_pref": {
             "nullable": true,
             "description": "Apply a local preference to routes received from this peer.",
+            "default": null,
             "type": "integer",
             "format": "uint32",
             "minimum": 0
           },
           "md5_auth_key": {
             "nullable": true,
             "description": "Use the given key for TCP-MD5 authentication with the peer.",
+            "default": null,
             "type": "string"
           },
           "min_ttl": {
             "nullable": true,
             "description": "Require messages from a peer have a minimum IP time to live field.",
+            "default": null,
             "type": "integer",
             "format": "uint8",
             "minimum": 0
           },
           "multi_exit_discriminator": {
             "nullable": true,
             "description": "Apply the provided multi-exit discriminator (MED) updates sent to the peer.",
+            "default": null,
             "type": "integer",
             "format": "uint32",
             "minimum": 0
@@ -467,13 +473,15 @@
           "remote_asn": {
             "nullable": true,
             "description": "Require that a peer has a specified ASN.",
+            "default": null,
             "type": "integer",
             "format": "uint32",
             "minimum": 0
           },
           "vlan_id": {
             "nullable": true,
             "description": "Associate a VLAN ID with a BGP peer session.",
+            "default": null,
             "type": "integer",
             "format": "uint16",
             "minimum": 0
@@ -1192,6 +1200,7 @@
           "vlan_id": {
             "nullable": true,
             "description": "The VLAN id associated with this route.",
+            "default": null,
             "type": "integer",
             "format": "uint16",
             "minimum": 0
@@ -1234,6 +1243,7 @@
           "vlan_id": {
             "nullable": true,
             "description": "The VLAN id (if any) associated with this address.",
+            "default": null,
             "type": "integer",
             "format": "uint16",
             "minimum": 0

diff --git a/openapi/nexus-internal.json b/openapi/nexus-internal.json
@@ -1567,6 +1567,7 @@
           "checker": {
             "nullable": true,
             "description": "Checker to apply to incoming messages.",
+            "default": null,
             "type": "string"
           },
           "originate": {
@@ -1579,6 +1580,7 @@
           "shaper": {
             "nullable": true,
             "description": "Shaper to apply to outgoing messages.",
+            "default": null,
             "type": "string"
           }
         },
@@ -1676,25 +1678,29 @@
           "local_pref": {
             "nullable": true,
             "description": "Apply a local preference to routes received from this peer.",
+            "default": null,
             "type": "integer",
             "format": "uint32",
             "minimum": 0
           },
           "md5_auth_key": {
             "nullable": true,
             "description": "Use the given key for TCP-MD5 authentication with the peer.",
+            "default": null,
             "type": "string"
           },
           "min_ttl": {
             "nullable": true,
             "description": "Require messages from a peer have a minimum IP time to live field.",
+            "default": null,
             "type": "integer",
             "format": "uint8",
             "minimum": 0
           },
           "multi_exit_discriminator": {
             "nullable": true,
             "description": "Apply the provided multi-exit discriminator (MED) updates sent to the peer.",
+            "default": null,
             "type": "integer",
             "format": "uint32",
             "minimum": 0
@@ -1706,13 +1712,15 @@
           "remote_asn": {
             "nullable": true,
             "description": "Require that a peer has a specified ASN.",
+            "default": null,
             "type": "integer",
             "format": "uint32",
             "minimum": 0
           },
           "vlan_id": {
             "nullable": true,
             "description": "Associate a VLAN ID with a BGP peer session.",
+            "default": null,
             "type": "integer",
             "format": "uint16",
             "minimum": 0
@@ -4345,6 +4353,7 @@
           "vlan_id": {
             "nullable": true,
             "description": "The VLAN id associated with this route.",
+            "default": null,
             "type": "integer",
             "format": "uint16",
             "minimum": 0
@@ -5003,6 +5012,7 @@
           "vlan_id": {
             "nullable": true,
             "description": "The VLAN id (if any) associated with this address.",
+            "default": null,
             "type": "integer",
             "format": "uint16",
             "minimum": 0

diff --git a/openapi/nexus.json b/openapi/nexus.json
@@ -9344,6 +9344,39 @@
           }
         ]
       },
+      "AuthzScope": {
+        "description": "Authorization scope for a timeseries.\n\nThis describes the level at which a user must be authorized to read data from a timeseries. For example, fleet-scoping means the data is only visible to an operator or fleet reader. Project-scoped, on the other hand, indicates that a user will see data limited to the projects on which they have read permissions.",
+        "oneOf": [
+          {
+            "description": "Timeseries data is limited to fleet readers.",
+            "type": "string",
+            "enum": [
+              "fleet"
+            ]
+          },
+          {
+            "description": "Timeseries data is limited to the authorized silo for a user.",
+            "type": "string",
+            "enum": [
+              "silo"
+            ]
+          },
+          {
+            "description": "Timeseries data is limited to the authorized projects for a user.",
+            "type": "string",
+            "enum": [
+              "project"
+            ]
+          },
+          {
+            "description": "The timeseries is viewable to all without limitation.",
+            "type": "string",
+            "enum": [
+              "viewable_to_all"
+            ]
+          }
+        ]
+      },
       "Baseboard": {
         "description": "Properties that uniquely identify an Oxide hardware component",
         "type": "object",
@@ -12646,6 +12679,9 @@
         "description": "The name and type information for a field of a timeseries schema.",
         "type": "object",
         "properties": {
+          "description": {
+            "type": "string"
+          },
           "field_type": {
             "$ref": "#/components/schemas/FieldType"
           },
@@ -12657,6 +12693,7 @@
           }
         },
         "required": [
+          "description",
           "field_type",
           "name",
           "source"
@@ -16686,6 +16723,7 @@
           "signing_keypair": {
             "nullable": true,
             "description": "request signing key pair",
+            "default": null,
             "allOf": [
               {
                 "$ref": "#/components/schemas/DerEncodedKeyPair"
@@ -18546,6 +18584,22 @@
           "points"
         ]
       },
+      "TimeseriesDescription": {
+        "description": "Text descriptions for the target and metric of a timeseries.",
+        "type": "object",
+        "properties": {
+          "metric": {
+            "type": "string"
+          },
+          "target": {
+            "type": "string"
+          }
+        },
+        "required": [
+          "metric",
+          "target"
+        ]
+      },
       "TimeseriesName": {
         "title": "The name of a timeseries",
         "description": "Names are constructed by concatenating the target and metric names with ':'. Target and metric names must be lowercase alphanumeric characters with '_' separating words.",
@@ -18569,13 +18623,19 @@
         "description": "The schema for a timeseries.\n\nThis includes the name of the timeseries, as well as the datum type of its metric and the schema for each field.",
         "type": "object",
         "properties": {
+          "authz_scope": {
+            "$ref": "#/components/schemas/AuthzScope"
+          },
           "created": {
             "type": "string",
             "format": "date-time"
           },
           "datum_type": {
             "$ref": "#/components/schemas/DatumType"
           },
+          "description": {
+            "$ref": "#/components/schemas/TimeseriesDescription"
+          },
           "field_schema": {
             "type": "array",
             "items": {
@@ -18585,13 +18645,25 @@
           },
           "timeseries_name": {
             "$ref": "#/components/schemas/TimeseriesName"
+          },
+          "units": {
+            "$ref": "#/components/schemas/Units"
+          },
+          "version": {
+            "type": "integer",
+            "format": "uint8",
+            "minimum": 1
           }
         },
         "required": [
+          "authz_scope",
           "created",
           "datum_type",
+          "description",
           "field_schema",
-          "timeseries_name"
+          "timeseries_name",
+          "units",
+          "version"
         ]
       },
       "TimeseriesSchemaResultsPage": {
@@ -18675,6 +18747,14 @@
           "items"
         ]
       },
+      "Units": {
+        "description": "Measurement units for timeseries samples.",
+        "type": "string",
+        "enum": [
+          "count",
+          "bytes"
+        ]
+      },
       "User": {
         "description": "View of a User",
         "type": "object",