Merge pull request 'Phase 9 R1: repo-ensure RPC + fleet-health fixes' (#107) from phase-9-r1a-repo-ensure-rpc into main

V2State::from_config built its peer fan-out list solely from
[[cluster.peers]], which by definition never includes the local
node. Result: hitting a given node's /api/v2/fleet directly always
omitted that node from its own fleet view, even when perfectly
healthy -- looked like "node X is missing" from the dashboard when
X just never queried itself.

Fix: synthesize a self PeerEntry from the node's own gossip bind
address and include it in the fan-out, same as any other peer.
peer_rpc_addr()'s existing +1 port convention resolves it to the
same bind_rpc_lan/bind_rpc_tailscale the daemon already listens on.

Co-Authored-By: Claude Sonnet 5 <[email protected]>

claw-store/src/daemon.rs

@@ -7,7 +7,7 @@ use crate::manifest::Manifest;
 use crate::snapshot;
 use crate::sync::{SyncQueue, drain_sync_queue};
 use crate::zfs::SystemZfs;
-use anyhow::Result;
+use anyhow::{Context, Result};
 use chrono::Utc;
 use sysinfo::{ProcessRefreshKind, RefreshKind, System};
 use tokio::time::{interval, Duration};
@@ -36,7 +36,14 @@ pub async fn run(cfg: Config, mut manifest: Manifest) -> Result<()> {
             let hot_dir = cfg.hot.path.clone();
             let hot_max_bytes = cfg.hot.max_gb.saturating_mul(1024 * 1024 * 1024);
             let blob_root = cluster_cfg.blob_store_root.clone();
-            match ClusterServices::start(
+            // Fail fast rather than degrade silently: a bind failure here is
+            // almost always a boot-time race against DHCP/network-online
+            // (the bind address isn't assigned to the interface yet). The
+            // systemd unit has `Restart=on-failure`; exiting lets it retry
+            // a few seconds later once the network is actually up, instead
+            // of leaving the daemon running indefinitely with no gossip,
+            // RPC, or Prometheus endpoint and no visible failure state.
+            let svc = ClusterServices::start(
                 cluster_cfg,
                 cfg.node.name.clone(),
                 hot_dir,
@@ -44,8 +51,7 @@ pub async fn run(cfg: Config, mut manifest: Manifest) -> Result<()> {
                 blob_root,
             )
             .await
-            {
+            .context("starting cluster services")?;
-                Ok(svc) => {
             tracing::info!(
                 rpc_enabled = svc.rpc_enabled(),
                 blob_store_enabled = svc.blob_store_enabled(),
@@ -54,12 +60,6 @@ pub async fn run(cfg: Config, mut manifest: Manifest) -> Result<()> {
             );
             Some(svc)
         }
-                Err(e) => {
-                    tracing::error!(error = %e, "cluster services failed to start; continuing without cluster");
-                    None
-                }
-            }
-        }
         None => {
             tracing::info!("no [cluster] section in config; running standalone");
             None

claw-store/src/serve_v2.rs

@@ -91,9 +91,30 @@ impl V2State {
             .join("aggregator-sessions.json");
         let sessions = SessionStore::load(sessions_path)
             .map_err(|e| anyhow::anyhow!("loading session store: {e}"))?;
+        // Bug fix 2026-07-31: the fleet view previously never included
+        // the node actually serving the dashboard — `cluster.peers` is
+        // by definition every *other* node, so hitting a given node's
+        // `/api/v2/fleet` directly silently dropped that node from its
+        // own view (looked like "node X is missing" from the UI, even
+        // though X was perfectly healthy — it just never queried
+        // itself). Fix: synthesize a self `PeerEntry` from our own
+        // gossip bind address and include it in the fan-out list, same
+        // as any other peer. `peer_rpc_addr` derives the RPC port from
+        // `lan_addr`/`tailscale_addr` via the fleet's +1 convention, so
+        // this resolves to the same `bind_rpc_lan`/`bind_rpc_tailscale`
+        // the daemon actually listens on.
+        let self_peer = PeerEntry {
+            name: cfg.node.name.clone(),
+            zone: cluster.zone.clone(),
+            lan_addr: cluster.bind_lan,
+            tailscale_addr: cluster.bind_tailscale,
+        };
+        let mut peers = cluster.peers.clone();
+        peers.push(self_peer);
+
         Ok(Self {
             aggregator_name: cfg.node.name.clone(),
-            peers: cluster.peers.clone(),
+            peers,
             client: std::sync::Arc::new(client),
             default_rpc_port_offset: 1,
             api_token: cfg.api_token.clone(),