All tests passing!

oxidecomputer · Nov 25, 2024 · 79652b6 · 79652b6
1 parent 3818b40
commit 79652b6
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 4 deletions.
diff --git a/openapi/crucible-control.json b/openapi/crucible-control.json
@@ -370,7 +370,7 @@
         "description": "Tracks client negotiation progress",
         "oneOf": [
           {
-            "description": "Initial state",
+            "description": "Initial state, waiting to hear `YesItsMe` from the client\n\nOnce this message is heard, transitions to either `WaitActive` (if `auto_promote` is `false`) or `WaitQuorum` (if `auto_promote` is `true`)",
             "type": "object",
             "properties": {
               "type": {

diff --git a/upstairs/src/client.rs b/upstairs/src/client.rs
@@ -1157,7 +1157,6 @@ impl DownstairsClient {
     /// error is at or after `Message::YouAreNowActive`.
     ///
     /// Returns a flag indicating how to proceed
-    #[must_use]
     pub(crate) fn continue_negotiation(
         &mut self,
         m: Message,
@@ -1626,6 +1625,20 @@ impl DownstairsClient {
                         *state = NegotiationState::WaitQuorum;
                         out = NegotiationResult::WaitQuorum;
                     }
+                    // Special case: if a downstairs is replaced while we're
+                    // still trying to go active, then we use the WaitQuorum
+                    // path instead of LiveRepair.
+                    ConnectionMode::Replaced
+                        if matches!(
+                            up_state,
+                            UpstairsState::Initializing
+                                | UpstairsState::GoActive(..)
+                        ) =>
+                    {
+                        *state = NegotiationState::WaitQuorum;
+                        out = NegotiationResult::WaitQuorum;
+                    }
+
                     ConnectionMode::Faulted | ConnectionMode::Replaced => {
                         *state = NegotiationState::LiveRepairReady;
                         out = NegotiationResult::LiveRepair;
@@ -1793,7 +1806,10 @@ impl DownstairsClient {
 #[serde(rename_all = "snake_case")]
 #[serde(tag = "type", content = "value")]
 pub enum NegotiationState {
-    /// Initial state
+    /// Initial state, waiting to hear `YesItsMe` from the client
+    ///
+    /// Once this message is heard, transitions to either `WaitActive` (if
+    /// `auto_promote` is `false`) or `WaitQuorum` (if `auto_promote` is `true`)
     Start {
         auto_promote: bool,
     },

diff --git a/upstairs/src/downstairs.rs b/upstairs/src/downstairs.rs
@@ -2595,6 +2595,14 @@ impl Downstairs {
                 continue;
             }
             match self.clients[client_id].state() {
+                // Replacement is allowed before starting, but not between
+                // activation and coming online
+                DsState::Connecting {
+                    state:
+                        NegotiationState::Start { .. }
+                        | NegotiationState::WaitActive,
+                    mode: ConnectionMode::New,
+                } => {}
                 DsState::Stopping(..)
                 | DsState::Connecting { .. }
                 | DsState::LiveRepair => {
@@ -2603,7 +2611,7 @@ impl Downstairs {
                         self.clients[client_id].state(),
                     )));
                 }
-                _ => {}
+                DsState::Active => {}
             }
         }