Merge branch 'eichhorl/key-alert' into 'master'

feat(consensus): CON-1041 ECDSA public key monitoring

Currently, if a subnet changes or loses its ECDSA key due to a bug, for instance after a replica upgrade, we will not be aware of it until the issue propagates to users.

With this MR we instead detect such a situation by inspecting the CUP of an ECDSA subnet. If its key differs from the one included in the previous CUP of the subnet, we increment an error metric and raise an alert.

In particular, by upgrading the ECDSA backup subnet first, this would allow us to be notified of certain problems in advance of upgrading the ECDSA signing subnet.

Note the [Design document](https://docs.google.com/document/d/1WBMkPHbUe8R4IWtuQsrOvlcKPzN14dwC6A4QvGUXyJc/edit) primarily talks about only monitoring the first CUP after an upgrade. Here we extend this idea to analogously compare every new CUP with the previous one.

One problem is monitoring key changes done in the last CUP before a restart. As the orchestrator typically shuts down immediately after inspecting the CUP, the error metric is reset before it can be scraped. This can be fixed by persisting the metric in the orchestrator's data directory, which will be done in a follow-up MR. 

See merge request dfinity-lab/public/ic!13013

rs/orchestrator/BUILD.bazel

@@ -43,6 +43,7 @@ rust_library(
         "//rs/registry/proto_data_provider",
         "//rs/registry/routing_table",
         "//rs/sys",
+        "//rs/types/ic00_types",
         "//rs/types/types",
         "//rs/utils",
         "@crate_index//:candid",
@@ -84,7 +85,10 @@ rust_test(
     crate = ":lib",
     deps = [
         "//rs/crypto/temp_crypto",
+        "//rs/crypto/test_utils/canister_threshold_sigs",
+        "//rs/crypto/test_utils/reproducible_rng",
         "//rs/registry/fake",
+        "//rs/test_utilities",
         "//rs/test_utilities/in_memory_logger",
         "//rs/test_utilities/logger",
         "@crate_index//:assert_cmd",

rs/orchestrator/Cargo.toml

@@ -23,6 +23,7 @@ ic-crypto-utils-threshold-sig = { path = "../crypto/utils/threshold_sig" }
 ic-crypto-sha = { path = "../crypto/sha" }
 ic-dashboard = { path = "./dashboard" }
 ic-http-utils = { path = "../http_utils" }
+ic-ic00-types = { path = "../types/ic00_types" }
 ic-image-upgrader = { path = "./image_upgrader" }
 ic-interfaces = { path = "../interfaces" }
 ic-interfaces-registry = { path = "../interfaces/registry" }
@@ -60,7 +61,10 @@ url = "2.1.1"
 [dev-dependencies]
 assert_cmd = "0.12"
 ic-crypto-temp-crypto = { path = "../crypto/temp_crypto" }
+ic-crypto-test-utils-canister-threshold-sigs = { path = "../crypto/test_utils/canister_threshold_sigs" }
+ic-crypto-test-utils-reproducible-rng = { path = "../crypto/test_utils/reproducible_rng" }
 ic-registry-client-fake = { path = "../registry/fake" }
+ic-test-utilities = { path = "../test_utilities" }
 ic-test-utilities-in-memory-logger = { path = "../test_utilities/in_memory_logger" }
 ic-test-utilities-logger = { path = "../test_utilities/logger" }
 mockall = "0.8.3"

rs/orchestrator/src/catch_up_package_provider.rs

@@ -310,7 +310,6 @@ impl CatchUpPackageProvider {
                 registry_version,
             ))?;
 
-        let unsigned = latest_cup.signature.signature.get_ref().0.is_empty();
         let height = Some(latest_cup.content.height());
         // We recreate the local registry CUP everytime to avoid incompatibility issues. Without
         // this recreation, we might run into the following problem: assume the orchestrator of
@@ -323,7 +322,7 @@ impl CatchUpPackageProvider {
         //
         // By re-creating the unsigned CUP every time we realize it's the newest one, we instead
         // recreate the CUP on all orchestrators of the version B before starting the replica.
-        if height > local_cup_height || height == local_cup_height && unsigned {
+        if height > local_cup_height || height == local_cup_height && latest_cup.is_unsigned() {
             self.persist_cup(&latest_cup_proto)?;
         }
         Ok((latest_cup_proto, latest_cup))

rs/orchestrator/src/metrics.rs

@@ -1,4 +1,4 @@
-use prometheus::{IntCounter, IntGauge, IntGaugeVec};
+use prometheus::{IntCounter, IntCounterVec, IntGauge, IntGaugeVec};
 use strum::IntoEnumIterator;
 use strum_macros::{EnumIter, IntoStaticStr};
 
@@ -15,6 +15,7 @@ pub struct OrchestratorMetrics {
     pub reboot_duration: IntGauge,
     pub orchestrator_info: IntGaugeVec,
     pub key_rotation_status: IntGaugeVec,
+    pub ecdsa_key_changed_errors: IntCounterVec,
 }
 
 #[derive(Copy, Clone, Debug, EnumIter, Eq, IntoStaticStr, PartialOrd, Ord, PartialEq)]
@@ -77,6 +78,11 @@ impl OrchestratorMetrics {
                 "The current key rotation status.",
                 &["status"],
             ),
+            ecdsa_key_changed_errors: metrics_registry.int_counter_vec(
+                "orchestrator_tecdsa_key_changed_errors_total",
+                "Critical error counter monitoring changed tECDSA public keys",
+                &["key_id"],
+            ),
         }
     }