fix: complete global router config and recoverable cluster init

src/cluster/cluster.rs

@@ -1,6 +1,8 @@
 //! Cluster: routes arrivals, performs the L0 / L1 / remote-RDMA fetch chain
 //! described in the design diagram, and bookkeeps the global meta store.
 
+use anyhow::Result;
+
 use crate::cluster::meta_store::MetaStore;
 use crate::config::{Config, ModelConfig};
 use crate::instance::instance::AdmittedRequest;
@@ -36,11 +38,8 @@ pub struct Cluster {
 }
 
 impl Cluster {
-    pub fn new(config: &Config, model: &ModelConfig) -> Self {
-        let total_instances = config
-            .cluster
-            .require_legacy_single_pool("Cluster::new")
-            .unwrap_or_else(|err| panic!("{err}"));
+    pub fn new(config: &Config, model: &ModelConfig) -> Result<Self> {
+        let total_instances = config.cluster.require_legacy_single_pool("Cluster::new")?;
         let mut instances = Vec::with_capacity(total_instances as usize);
         for id in 0..total_instances {
             instances.push(Instance::new(
@@ -52,7 +51,7 @@ impl Cluster {
         }
         let meta_store = MetaStore::new(config.cluster.meta_store.ttl_seconds);
         let router = router::build(config, config.sim.seed);
-        Self {
+        Ok(Self {
             instances,
             meta_store,
             router,
@@ -63,7 +62,7 @@ impl Cluster {
                 &config.calibration,
                 model.kv_block_bytes(),
             ),
-        }
+        })
     }
 
     /// Route + admit a request. Returns the chosen instance plus rich
@@ -262,7 +261,7 @@ mod tests {
     #[test]
     fn l1_ready_at_includes_dram_and_transform_overhead() {
         let cfg = test_config(RouterMode::EstimatedTtft);
-        let mut cluster = Cluster::new(&cfg, &cfg.model);
+        let mut cluster = Cluster::new(&cfg, &cfg.model).unwrap();
         let req = RequestRecord {
             req_id: 1,
             chat_id: 0,
@@ -300,17 +299,10 @@ mod tests {
             num_instances: 2,
         }];
 
-        let panic = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| {
-            Cluster::new(&cfg, &cfg.model);
-        }))
-        .expect_err("bucketed Cluster::new should panic");
-
-        let msg = panic
-            .downcast_ref::<String>()
-            .cloned()
-            .or_else(|| panic.downcast_ref::<&str>().map(|s| (*s).to_string()))
-            .expect("panic payload should be a string");
-        assert!(msg.contains("Cluster::new"));
-        assert!(msg.contains("cluster.buckets"));
+        let result = Cluster::new(&cfg, &cfg.model);
+        assert!(result.is_err(), "bucketed Cluster::new should fail");
+        let err = result.err().unwrap();
+        assert!(err.to_string().contains("Cluster::new"));
+        assert!(err.to_string().contains("cluster.buckets"));
     }
 }