[nexus] allow abbreviating ereport 'class' as 'k' (#8804)

This was one of @cbiffle's ideas to save a few bytes of CBO. Since we
expect to encode a "class" key for every ereport, shortening the fields
that are included in every message has a meaningful impact on how much
we can buffer. Also, loss records (which are generated when the SP
restarts or when ereports don't fit in the buffer) lack a class string,
because the presence of a top-level "lost" key can be used to identify
them. So, we should handle that here, as well.

Also includes some SP sim improvements for testing this.

Author: hawkw

gateway-test-utils/configs/sp_sim_config.test.toml

@@ -55,7 +55,7 @@ restart_id = "0d3e464a-666e-4687-976f-90e31238be8b"
 task_name = "task_thermal_server"
 task_gen = 1
 uptime = 1235
-class = "oxide.sidecar.thermal.sensor_read_error"
+k = "oxide.sidecar.thermal.sensor_read_error"
 sensor = { id = "dev-1", device = "fake-tmp-sensor", location = "South", presence = "Failed" }
 error = "DeviceError"
 
@@ -209,14 +209,14 @@ restart_id = "af1ebf85-36ba-4c31-bbec-b9825d6d9d8b"
 task_name = "task_apollo_server"
 task_gen = 13
 uptime = 1233
-class = "gov.nasa.apollo.o2_tanks.stir.begin"
+k = "gov.nasa.apollo.o2_tanks.stir.begin"
 message = "stirring the tanks"
 
 [[simulated_sps.gimlet.ereport_config.ereports]]
 task_name = "drv_ae35_server"
 task_gen = 1
 uptime = 1234
-class = "io.discovery.ae35.fault"
+k = "io.discovery.ae35.fault"
 message = "i've just picked up a fault in the AE-35 unit"
 de = { scheme = "fmd", authority = { product-id = "HAL-9000-series computer", server-id = "HAL 9000"}, mod-name = "ae35-diagnosis" }
 hours_to_failure = 72
@@ -225,22 +225,22 @@ hours_to_failure = 72
 task_name = "task_apollo_server"
 task_gen = 13
 uptime = 1237
-class = "gov.nasa.apollo.fault"
+k = "gov.nasa.apollo.fault"
 message = "houston, we have a problem"
 crew = ["Lovell", "Swigert", "Haise"]
 
 [[simulated_sps.gimlet.ereport_config.ereports]]
 task_name = "drv_thingy_server"
 task_gen = 2
 uptime = 1240
-class = "flagrant_error"
+k = "flagrant_error"
 computer = false
 
 [[simulated_sps.gimlet.ereport_config.ereports]]
 task_name = "task_latex_server"
 task_gen = 1
 uptime = 1245
-class = "overfull_hbox"
+k = "overfull_hbox"
 badness  = 10000
 
 [[simulated_sps.gimlet]]

gateway/tests/integration_tests/ereports.rs

@@ -72,6 +72,20 @@ mod sled0 {
         "af1ebf85-36ba-4c31-bbec-b9825d6d9d8b".parse().expect("is a valid UUID")
     });
 
+    def_ereport! {
+        LOSS: {
+            "baseboard_part_number": "SimGimletSp",
+            "baseboard_serial_number": "SimGimlet00",
+            "hubris_archive_id": "ffffffff",
+            "hubris_version": "0.0.2",
+            "hubris_task_name": "packrat",
+            "hubris_task_gen": 0,
+            "hubris_uptime_ms": 666,
+            "ereport_message_version": 0,
+            "lost": null,
+        }
+    }
+
     def_ereport! {
         EREPORT_1: {
             "baseboard_part_number": "SimGimletSp",
@@ -82,7 +96,7 @@ mod sled0 {
             "hubris_task_gen": 13,
             "hubris_uptime_ms": 1233,
             "ereport_message_version": 0,
-            "class": "gov.nasa.apollo.o2_tanks.stir.begin",
+            "k": "gov.nasa.apollo.o2_tanks.stir.begin",
             "message": "stirring the tanks",
         }
     }
@@ -96,7 +110,7 @@ mod sled0 {
             "hubris_task_gen": 1,
             "hubris_uptime_ms": 1234,
             "ereport_message_version": 0,
-            "class": "io.discovery.ae35.fault",
+            "k": "io.discovery.ae35.fault",
             "message": "i've just picked up a fault in the AE-35 unit",
             "de": {
                 "scheme": "fmd",
@@ -119,7 +133,7 @@ mod sled0 {
             "hubris_task_gen": 13,
             "hubris_uptime_ms": 1237,
             "ereport_message_version": 0,
-            "class": "gov.nasa.apollo.fault",
+            "k": "gov.nasa.apollo.fault",
             "message": "houston, we have a problem",
             "crew": [
                 "Lovell",
@@ -139,7 +153,7 @@ mod sled0 {
             "hubris_task_gen": 2,
             "hubris_uptime_ms": 1240,
             "ereport_message_version": 0,
-            "class": "flagrant_error",
+            "k": "flagrant_error",
             "computer": false,
         }
     }
@@ -154,7 +168,7 @@ mod sled0 {
             "hubris_task_gen": 1,
             "hubris_uptime_ms": 1245,
             "ereport_message_version": 0,
-            "class": "overfull_hbox",
+            "k": "overfull_hbox",
             "badness": 10000,
         }
     }
@@ -166,6 +180,20 @@ mod sled1 {
         "55e30cc7-a109-492f-aca9-735ed725df3c".parse().expect("is a valid UUID")
     });
 
+    def_ereport! {
+        LOSS: {
+            "baseboard_part_number": "SimGimletSp",
+            "baseboard_serial_number": "SimGimlet01",
+            "hubris_archive_id": "ffffffff",
+            "hubris_version": "0.0.2",
+            "hubris_task_name": "packrat",
+            "hubris_task_gen": 0,
+            "hubris_uptime_ms": 666,
+            "ereport_message_version": 0,
+            "lost": null,
+        }
+    }
+
     def_ereport! {
         EREPORT_1: {
             "baseboard_part_number": "SimGimletSp",
@@ -233,9 +261,14 @@ async fn ereports_basic() {
 
     assert_eq!(restart_id.as_untyped_uuid(), &*sled1::RESTART_0);
     let reports = reports.items;
-    assert_eq!(reports.len(), 1, "expected 1 ereport, found: {:#?}", reports);
+    assert_eq!(reports.len(), 2, "expected 2 ereports, found: {:#?}", reports);
+
     let report = &reports[0];
     assert_eq!(report.ena, ereport_types::Ena(1));
+    assert_eq!(report.data, *sled1::LOSS);
+
+    let report = &reports[1];
+    assert_eq!(report.ena, ereport_types::Ena(2));
     assert_eq!(report.data, *sled1::EREPORT_1);
 
     testctx.teardown().await;
@@ -254,21 +287,26 @@ async fn ereports_limit() {
             restart_id: Uuid::new_v4(),
             start_ena: 0,
             committed_ena: None,
-            limit: 2
+            limit: 3
         }
         .response(client)
         .await
     );
 
     assert_eq!(restart_id.as_untyped_uuid(), &*sled0::RESTART_0);
     let reports = reports.items;
-    assert_eq!(reports.len(), 2, "expected 2 ereports, found: {:#?}", reports);
+    assert_eq!(reports.len(), 3, "expected 3 ereports, found: {:#?}", reports);
+
     let report = &reports[0];
     assert_eq!(report.ena, ereport_types::Ena(1));
-    assert_eq!(report.data, *sled0::EREPORT_1);
+    assert_eq!(report.data, *sled0::LOSS);
 
     let report = &reports[1];
     assert_eq!(report.ena, ereport_types::Ena(2));
+    assert_eq!(report.data, *sled0::EREPORT_1);
+
+    let report = &reports[2];
+    assert_eq!(report.ena, ereport_types::Ena(3));
     assert_eq!(report.data, *sled0::EREPORT_2);
 
     let ereport_types::Ereports { restart_id, reports } = dbg!(
@@ -288,11 +326,11 @@ async fn ereports_limit() {
     assert_eq!(reports.len(), 2, "expected 2 ereports, found: {:#?}", reports);
     let report = &reports[0];
     assert_eq!(report.ena, ereport_types::Ena(3));
-    assert_eq!(report.data, *sled0::EREPORT_3);
+    assert_eq!(report.data, *sled0::EREPORT_2);
 
     let report = &reports[1];
     assert_eq!(report.ena, ereport_types::Ena(4));
-    assert_eq!(report.data, *sled0::EREPORT_4);
+    assert_eq!(report.data, *sled0::EREPORT_3);
 
     testctx.teardown().await;
 }
@@ -323,11 +361,11 @@ async fn ereports_commit() {
     assert_eq!(reports.len(), 2, "expected 2 ereports, found: {:#?}", reports);
     let report = &reports[0];
     assert_eq!(report.ena, ereport_types::Ena(1));
-    assert_eq!(report.data, *sled0::EREPORT_1);
+    assert_eq!(report.data, *sled0::LOSS);
 
     let report = &reports[1];
     assert_eq!(report.ena, ereport_types::Ena(2));
-    assert_eq!(report.data, *sled0::EREPORT_2);
+    assert_eq!(report.data, *sled0::EREPORT_1);
 
     // Now, send a request with a committed ENA *and* a matching restart ID.
     let ereport_types::Ereports { restart_id, reports } = dbg!(
@@ -347,11 +385,11 @@ async fn ereports_commit() {
     assert_eq!(reports.len(), 2, "expected 2 ereports, found: {:#?}", reports);
     let report = &reports[0];
     assert_eq!(report.ena, ereport_types::Ena(3));
-    assert_eq!(report.data, *sled0::EREPORT_3);
+    assert_eq!(report.data, *sled0::EREPORT_2);
 
     let report = &reports[1];
     assert_eq!(report.ena, ereport_types::Ena(4));
-    assert_eq!(report.data, *sled0::EREPORT_4);
+    assert_eq!(report.data, *sled0::EREPORT_3);
 
     // Even if the start ENA of a subsequent request is 0, we shouldn't see any
     // ereports with ENAs lower than the committed ENA.
@@ -369,17 +407,21 @@ async fn ereports_commit() {
 
     assert_eq!(restart_id.as_untyped_uuid(), &*sled0::RESTART_0);
     let reports = reports.items;
-    assert_eq!(reports.len(), 3, "expected 3 ereports, found: {:#?}", reports);
+    assert_eq!(reports.len(), 4, "expected 3 ereports, found: {:#?}", reports);
     let report = &reports[0];
     assert_eq!(report.ena, ereport_types::Ena(3));
-    assert_eq!(report.data, *sled0::EREPORT_3);
+    assert_eq!(report.data, *sled0::EREPORT_2);
 
     let report = &reports[1];
     assert_eq!(report.ena, ereport_types::Ena(4));
-    assert_eq!(report.data, *sled0::EREPORT_4);
+    assert_eq!(report.data, *sled0::EREPORT_3);
 
     let report = &reports[2];
     assert_eq!(report.ena, ereport_types::Ena(5));
+    assert_eq!(report.data, *sled0::EREPORT_4);
+
+    let report = &reports[3];
+    assert_eq!(report.ena, ereport_types::Ena(6));
     assert_eq!(report.data, *sled0::EREPORT_5);
 
     testctx.teardown().await;