fix(cuprate): enable fast_sync and raise DB cache — 45% CPU on amishparadise

The default manifest baked in the exact broken config found on
amishparadise: no fast_sync (defaults false, forcing full ring-sig/RandomX
verification on every block) and target_max_memory capped at ~2.8GiB,
which starved cuprated's DB cache into constant eviction/flush (595GB/24h
of block I/O on a node just appending ~2MB blocks every 2 minutes). A
reference node with fast_sync = true and an 8GiB cache ran at 2.8% CPU
at the same chain height and block rate.

Set fast_sync = true and target_max_memory = 8GiB to match the healthy
reference config, and raise resources.memory_limit from 4Gi to 10Gi so
the container still has headroom above the larger cache.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RR7jRaicvqsJaqQQ92jpPQ

apps/cuprate/manifest.yml

@@ -45,7 +45,12 @@ app:
 
   resources:
     cpu_limit: 0
-    memory_limit: 4Gi
+    # Raised from 4Gi alongside target_max_memory below (see files[] comment)
+    # — 2026-09-03 incident on amishparadise: a 4Gi/3GB-cache config starved
+    # cuprated's DB cache into constant eviction/flush, driving 45% sustained
+    # CPU and ~595GB/24h of block I/O on a fully-synced node. 10Gi leaves
+    # headroom above the 8GiB cache for the process itself.
+    memory_limit: 10Gi
     disk_limit: 300Gi
 
   security:
@@ -103,11 +108,24 @@ app:
   # Settings that need to differ from cuprated's own documented defaults
   # (verified against `cuprated --generate-config` and `--dry-run` locally,
   # 2026-08-21):
+  #   - fast_sync: cuprated's own default is false, which performs full
+  #     cryptographic verification (ring signatures + RandomX PoW) on every
+  #     incoming block instead of trusting checkpointed history. Root-caused
+  #     2026-09-03 as the dominant cause of a sustained 45% CPU node on
+  #     amishparadise, vs. 2.8% on a
+  #     reference node with fast_sync = true — same chain height, same
+  #     block rate. Set explicitly rather than relying on the binary
+  #     default so fresh deploys don't silently regress into full-verify.
   #   - target_max_memory: cuprated's own default auto-detects total *host*
   #     RAM via sysinfo, which inside a memory-limited container would let
   #     it size caches far past what resources.memory_limit above actually
   #     grants — same class of problem bitcoin-knots' -dbcache sizing
-  #     comment addresses. Set explicitly, comfortably under the 4Gi limit.
+  #     comment addresses. Set explicitly, comfortably under the 10Gi limit.
+  #     Previously 3000000000 (~2.8GiB); that starved the DB cache and
+  #     forced constant eviction/flush (595GB/24h block I/O on a node just
+  #     appending ~2MB blocks every 2 minutes) — raised to 8GiB, matching
+  #     the healthy reference node (ssmithx@archy-dev-pa), and
+  #     resources.memory_limit above raised in step to keep headroom above it.
   #   - rpc.restricted.enable: cuprated ships this off by default; flip on
   #     so the auth:none host port above actually serves something instead
   #     of refusing every connection. port stays at its documented default
@@ -142,7 +160,8 @@ app:
     - path: /var/lib/archipelago/cuprate/Cuprated.toml
       content: |
         network = "Mainnet"
-        target_max_memory = 3000000000
+        fast_sync = true
+        target_max_memory = 8589934592
 
         [rpc.restricted]
         enable = true