fix: handle pruned RPC state in SubQuery node

Substrate nodes prune historical state by default (~256 blocks).
When SubQuery restarts and tries to fetch runtime version for old
blocks, it crashes with "State already discarded". This patch adds
a fallback to current runtime version when historical state is
unavailable, preventing crash loops on pruned RPC endpoints.

Also removes --pull from deploy workflow to avoid unnecessary
full rebuilds on every push.

.github/workflows/deploy-vps.yml

@@ -15,5 +15,5 @@ jobs:
           script: |
             cd /opt/subquery
             git pull origin main
-            docker compose -f docker-compose.prod.yml build --pull
+            docker compose -f docker-compose.prod.yml build
             docker compose -f docker-compose.prod.yml up -d

docker/Dockerfile.node

@@ -41,4 +41,9 @@ RUN node -e " \
   fs.writeFileSync(p, JSON.stringify(pkg, null, 2)); \
 "
 
+# Patch: handle pruned blockchain state gracefully.
+# Without this, SubQuery crashes when RPC nodes have pruned old block state.
+COPY docker/patches/pruned-state-fallback.js /tmp/pruned-state-fallback.js
+RUN node /tmp/pruned-state-fallback.js && rm /tmp/pruned-state-fallback.js
+
 ENTRYPOINT ["/app/node_modules/.bin/subql-node"]

docker/patches/pruned-state-fallback.js

@@ -0,0 +1,54 @@
+/**
+ * Patch @subql/node to handle pruned blockchain state.
+ *
+ * Problem: Substrate nodes prune historical state by default (~256 blocks).
+ * When SubQuery restarts and tries to fetch runtime version for old blocks,
+ * state_getRuntimeVersion fails with "State already discarded" and the node
+ * enters a crash loop.
+ *
+ * Fix: When getRuntimeVersion(blockHash) fails due to pruned state,
+ * fall back to getRuntimeVersion() (current runtime version).
+ * This is safe for chains with infrequent spec upgrades.
+ */
+const fs = require("fs");
+
+// --- Patch 1: base-runtime.service.js ---
+// Handles getSpecFromApi() and getRuntimeVersion() which call
+// this.api.rpc.state.getRuntimeVersion(parentBlockHash)
+const runtimeFile =
+  "/app/node_modules/@subql/node/dist/indexer/runtime/base-runtime.service.js";
+let runtimeCode = fs.readFileSync(runtimeFile, "utf8");
+
+const runtimeMatches = (
+  runtimeCode.match(
+    /await this\.api\.rpc\.state\.getRuntimeVersion\(\w+\)/g
+  ) || []
+).length;
+
+runtimeCode = runtimeCode.replace(
+  /await this\.api\.rpc\.state\.getRuntimeVersion\((\w+)\)/g,
+  "await this.api.rpc.state.getRuntimeVersion($1).catch(() => this.api.rpc.state.getRuntimeVersion())"
+);
+fs.writeFileSync(runtimeFile, runtimeCode);
+console.log(`Patched base-runtime.service.js (${runtimeMatches} call sites)`);
+
+// --- Patch 2: utils/substrate.js ---
+// Handles fetchRuntimeVersionRange() which calls
+// api.rpc.state.getRuntimeVersion(hash).catch((e) => { throw ... })
+const utilsFile =
+  "/app/node_modules/@subql/node/dist/utils/substrate.js";
+let utilsCode = fs.readFileSync(utilsFile, "utf8");
+
+const utilsBefore = utilsCode.includes("getRuntimeVersion(hash).catch((e)");
+
+// Insert a fallback .catch before the existing error handler:
+// Original: getRuntimeVersion(hash).catch((e) => { throw... })
+// Patched:  getRuntimeVersion(hash).catch(() => getRuntimeVersion()).catch((e) => { throw... })
+utilsCode = utilsCode.replace(
+  /api\.rpc\.state\.getRuntimeVersion\(hash\)\.catch\(\(e\)/g,
+  "api.rpc.state.getRuntimeVersion(hash).catch(() => api.rpc.state.getRuntimeVersion()).catch((e)"
+);
+fs.writeFileSync(utilsFile, utilsCode);
+console.log(`Patched utils/substrate.js (had target: ${utilsBefore})`);
+
+console.log("All pruned-state patches applied.");