chore: Simplify and fix Snapshot metrics

crates/store/src/state/lifecycle.rs

@@ -3,7 +3,6 @@
 use std::num::NonZeroUsize;
 use std::path::Path;
 use std::sync::Arc;
-use std::sync::atomic::AtomicUsize;
 
 use arc_swap::ArcSwap;
 use miden_node_tracing::spawn::spawn_blocking_in_current_span;
@@ -32,15 +31,7 @@ use crate::state::loader::{
     verify_tree_consistency,
 };
 use crate::state::writer::{WriteRequest, WriteWorker, WriterTask};
-use crate::state::{
-    BlockCache,
-    BlockWriter,
-    ProofCache,
-    ProofWriter,
-    SnapshotGuard,
-    State,
-    StateSnapshot,
-};
+use crate::state::{BlockCache, BlockWriter, ProofCache, ProofWriter, State, StateSnapshot};
 use crate::{COMPONENT, DataDirectory, DatabaseOptions};
 
 /// Awaits a spawned load task, forwarding its result.
@@ -291,9 +282,6 @@ impl State {
         let block_cache = BlockCache::new(BLOCK_CACHE_CAPACITY);
         let proof_cache = ProofCache::new(PROOF_CACHE_CAPACITY);
 
-        // Shared counter of live snapshot generations, for observability.
-        let snapshots_live = Arc::new(AtomicUsize::new(0));
-
         // Create the initial snapshot from reader views of the just-loaded trees.
         let initial_snapshot = Arc::new(StateSnapshot::new(
             nullifier_tree
@@ -304,7 +292,6 @@ impl State {
             forest
                 .reader()
                 .map_err(|e| StateInitializationError::AccountStateForestIoError(e.as_report()))?,
-            SnapshotGuard::new(Arc::clone(&snapshots_live), latest_block_num),
         ));
         let latest_snapshot = Arc::new(ArcSwap::from(initial_snapshot));
 
@@ -324,7 +311,6 @@ impl State {
             account_tree,
             blockchain,
             forest,
-            snapshots_live,
             apply_block_thread_priority,
         );
         let state = Self {

crates/store/src/state/mod.rs

@@ -21,8 +21,8 @@ pub use lifecycle::LoadedState;
 use miden_protocol::block::BlockNumber;
 pub use replica::{BlockCache, BlockNotification, ProofCache, ProofNotification};
 use tokio::sync::watch;
+use view::StateSnapshot;
 pub use view::{ScopedBlockNum, ScopedBlockRange, StateView, StateWitnesses, TransactionInputs};
-use view::{SnapshotGuard, StateSnapshot};
 pub use writer::{BlockWriter, ProofWriter, WriterTask};
 
 use crate::blocks::BlockStore;

crates/store/src/state/view/mod.rs

@@ -27,12 +27,7 @@ mod scoped;
 pub use scoped::{ScopedBlockNum, ScopedBlockRange};
 
 mod snapshot;
-pub(in crate::state) use snapshot::{
-    PublishedGenerations,
-    SNAPSHOTS_LIVE_WARN_THRESHOLD,
-    SnapshotGuard,
-    StateSnapshot,
-};
+pub(in crate::state) use snapshot::{PublishedGenerations, StateSnapshot};
 
 mod account;
 mod block;
@@ -53,8 +48,7 @@ pub use transaction_inputs::TransactionInputs;
 ///
 /// Obtained from [`State::view`]; create one per request and drop it when the request completes.
 /// Holding a view pins a snapshot generation (and thereby the `RocksDB` snapshots backing the
-/// trees), so it must not be stored in long-lived structs; leaked or slow readers are reported by
-/// the store's snapshot-lifetime warnings.
+/// trees), so it must not be stored in long-lived structs.
 ///
 /// Reads that are technically not block-scoped (for example, immutable content-addressed data)
 /// also live here so that every read path flows through one type.
@@ -97,9 +91,7 @@ impl State {
     ///
     /// Work in the closure should be kept to low-complexity compute over the view, ideally with no
     /// I/O and no other `.await` points. Anything slower holds the pinned snapshot, and therefore
-    /// its underlying `RocksDB` snapshot, for as long as it runs. The snapshot's lifetime is logged
-    /// as a warning if held too long, but that is a backstop, not a substitute for keeping closures
-    /// short.
+    /// its underlying `RocksDB` snapshot, for as long as it runs.
     pub async fn with_view<R>(&self, f: impl AsyncFnOnce(&StateView) -> R) -> R {
         let view = self.view();
         f(&view).await