feat: improve operation lifecycle logging for CI debugging

Add real-time visibility into network operations to help debug
flaky CI test failures:

- Add --nocapture to CI test command for unbuffered output
- Enhance timeout logs with elapsed time and TTL info
- Add elapsed_ms to PUT and Connect completion logs
- Make Transaction::elapsed() public for logging access

This will show operation timing directly in CI logs:
  [INFO] Connect operation started tx=... gateway=...
  [INFO] Transaction timed out tx=... elapsed_ms=60012 ttl_ms=60000
  [INFO] PUT operation completed tx=... elapsed_ms=1234

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

Author: sanity

.github/workflows/ci.yml

@@ -129,7 +129,8 @@ jobs:
       - name: Test
         # Limit test threads to reduce resource contention on high-core-count runners
         # Without this, 64+ parallel tests cause timing-sensitive network tests to fail
-        run: cargo test --workspace --no-default-features --features trace,websocket,redb -- --test-threads=8
+        # --nocapture shows logs in real-time instead of buffering until test completion
+        run: cargo test --workspace --no-default-features --features trace,websocket,redb -- --test-threads=8 --nocapture
 
   six_peer_regression:
     name: six-peer-regression

crates/core/src/message.rs

@@ -98,7 +98,8 @@ impl Transaction {
         self.id.0.to_le_bytes()
     }
 
-    fn elapsed(&self) -> Duration {
+    /// Returns the elapsed time since this transaction was created.
+    pub fn elapsed(&self) -> Duration {
         let current_unix_epoch_ts = SystemTime::now()
             .duration_since(SystemTime::UNIX_EPOCH)
             .expect("now should be always be later than unix epoch")

crates/core/src/node/op_state_manager.rs

@@ -812,7 +812,13 @@ async fn garbage_cleanup_task<ER: NetEventRegister>(
                     } else {
                         ops.under_progress.remove(&tx);
                         ops.completed.remove(&tx);
-                        tracing::debug!("Transaction timed out: {tx}");
+                        tracing::info!(
+                            tx = %tx,
+                            tx_type = ?tx.transaction_type(),
+                            elapsed_ms = tx.elapsed().as_millis(),
+                            ttl_ms = crate::config::OPERATION_TTL.as_millis(),
+                            "Transaction timed out"
+                        );
 
                         // Check if this is a child operation and propagate timeout to parent
                         if let Some(parent_tx) = sub_op_tracker.get_parent(tx) {
@@ -857,7 +863,13 @@ async fn garbage_cleanup_task<ER: NetEventRegister>(
                         TransactionType::Update => ops.update.remove(&tx).is_some(),
                     };
                     if removed {
-                        tracing::debug!("Transaction timed out: {tx}");
+                        tracing::info!(
+                            tx = %tx,
+                            tx_type = ?tx.transaction_type(),
+                            elapsed_ms = tx.elapsed().as_millis(),
+                            ttl_ms = crate::config::OPERATION_TTL.as_millis(),
+                            "Transaction timed out"
+                        );
 
                         // Check if this is a child operation and propagate timeout to parent
                         if let Some(parent_tx) = sub_op_tracker.get_parent(tx) {

crates/core/src/operations/connect.rs

@@ -678,6 +678,11 @@ impl ConnectOp {
                     self.recency.remove(&new_acceptor.peer);
                 }
                 if result.satisfied {
+                    tracing::info!(
+                        tx = %self.id,
+                        elapsed_ms = self.id.elapsed().as_millis(),
+                        "Connect operation completed"
+                    );
                     self.state = Some(ConnectState::Completed);
                 }
                 Some(result)
@@ -975,11 +980,13 @@ impl Operation for ConnectOp {
                                         tracing::info!(
                                             %peer_id,
                                             tx=%self.id,
+                                            elapsed_ms = self.id.elapsed().as_millis(),
                                             "connect joined peer"
                                         );
                                     } else {
                                         tracing::warn!(
                                             tx=%self.id,
+                                            elapsed_ms = self.id.elapsed().as_millis(),
                                             "connect ConnectPeer failed"
                                         );
                                     }

crates/core/src/operations/put.rs

@@ -341,6 +341,7 @@ impl Operation for PutOp {
                         tracing::info!(
                             tx = %id,
                             %key,
+                            elapsed_ms = id.elapsed().as_millis(),
                             "PUT operation completed successfully"
                         );