add automatic heap snapshots for high-memory cli processes (#20788)

4 files changed, 66 insertions, 0 deletions

diff --git a/packages/opencode/src/cli/cmd/tui/worker.ts b/packages/opencode/src/cli/cmd/tui/worker.ts
index a83645d89..643676e34 100644
--- a/packages/opencode/src/cli/cmd/tui/worker.ts
+++ b/packages/opencode/src/cli/cmd/tui/worker.ts
@@ -13,6 +13,7 @@ import { Flag } from "@/flag/flag"
 import { setTimeout as sleep } from "node:timers/promises"
 import { writeHeapSnapshot } from "node:v8"
 import { WorkspaceID } from "@/control-plane/schema"
+import { Heap } from "@/cli/heap"
 
 await Log.init({
   print: process.argv.includes("--print-logs"),
@@ -23,6 +24,8 @@ await Log.init({
   })(),
 })
 
+Heap.start()
+
 process.on("unhandledRejection", (e) => {
   Log.Default.error("rejection", {
     e: e instanceof Error ? e.message : e,
diff --git a/packages/opencode/src/cli/heap.ts b/packages/opencode/src/cli/heap.ts
new file mode 100644
index 000000000..bb5a3d093
--- /dev/null
+++ b/packages/opencode/src/cli/heap.ts
@@ -0,0 +1,59 @@
+import path from "path"
+import { writeHeapSnapshot } from "node:v8"
+import { Flag } from "@/flag/flag"
+import { Global } from "@/global"
+import { Log } from "@/util/log"
+
+const log = Log.create({ service: "heap" })
+const MINUTE = 60_000
+const LIMIT = 2 * 1024 * 1024 * 1024
+
+export namespace Heap {
+  let timer: Timer | undefined
+  let lock = false
+  let armed = true
+
+  export function start() {
+    if (!Flag.OPENCODE_AUTO_HEAP_SNAPSHOT) return
+    if (timer) return
+
+    const run = async () => {
+      if (lock) return
+
+      const stat = process.memoryUsage()
+      if (stat.rss <= LIMIT) {
+        armed = true
+        return
+      }
+      if (!armed) return
+
+      lock = true
+      armed = false
+      const file = path.join(
+        Global.Path.log,
+        `heap-${process.pid}-${new Date().toISOString().replace(/[:.]/g, "")}.heapsnapshot`,
+      )
+      log.warn("heap usage exceeded limit", {
+        rss: stat.rss,
+        heap: stat.heapUsed,
+        file,
+      })
+
+      await Promise.resolve()
+        .then(() => writeHeapSnapshot(file))
+        .catch((err) => {
+          log.error("failed to write heap snapshot", {
+            error: err instanceof Error ? err.message : String(err),
+            file,
+          })
+        })
+
+      lock = false
+    }
+
+    timer = setInterval(() => {
+      void run()
+    }, MINUTE)
+    timer.unref?.()
+  }
+}
diff --git a/packages/opencode/src/flag/flag.ts b/packages/opencode/src/flag/flag.ts
index 27190f2eb..1ac52dd17 100644
--- a/packages/opencode/src/flag/flag.ts
+++ b/packages/opencode/src/flag/flag.ts
@@ -12,6 +12,7 @@ function falsy(key: string) {
 
 export namespace Flag {
   export const OPENCODE_AUTO_SHARE = truthy("OPENCODE_AUTO_SHARE")
+  export const OPENCODE_AUTO_HEAP_SNAPSHOT = truthy("OPENCODE_AUTO_HEAP_SNAPSHOT")
   export const OPENCODE_GIT_BASH_PATH = process.env["OPENCODE_GIT_BASH_PATH"]
   export const OPENCODE_CONFIG = process.env["OPENCODE_CONFIG"]
   export declare const OPENCODE_PURE: boolean
diff --git a/packages/opencode/src/index.ts b/packages/opencode/src/index.ts
index bb14e0588..1fa027abf 100644
--- a/packages/opencode/src/index.ts
+++ b/packages/opencode/src/index.ts
@@ -35,6 +35,7 @@ import { JsonMigration } from "./storage/json-migration"
 import { Database } from "./storage/db"
 import { errorMessage } from "./util/error"
 import { PluginCommand } from "./cli/cmd/plug"
+import { Heap } from "./cli/heap"
 
 process.on("unhandledRejection", (e) => {
   Log.Default.error("rejection", {
@@ -96,6 +97,8 @@ const cli = yargs(args)
       })(),
     })
 
+    Heap.start()
+
     process.env.AGENT = "1"
     process.env.OPENCODE = "1"
     process.env.OPENCODE_PID = String(process.pid)