Diagnostics (#101)

* Update all the deps

* Wait for the daemon to shutdown

* comment

* comment again

package.json

@@ -28,23 +28,23 @@
     "@actions/core": "^1.11.1",
     "@actions/exec": "^1.1.1",
     "detsys-ts": "github:DeterminateSystems/detsys-ts",
-    "got": "^14.4.4",
+    "got": "^14.4.5",
     "tail": "^2.2.6"
   },
   "devDependencies": {
     "@trivago/prettier-plugin-sort-imports": "^4.3.0",
-    "@types/node": "^20.17.6",
+    "@types/node": "^20.17.14",
     "@types/tail": "^2.2.3",
     "@types/uuid": "^9.0.8",
     "@typescript-eslint/eslint-plugin": "^7.18.0",
     "@vercel/ncc": "^0.38.3",
     "eslint": "^8.57.1",
-    "eslint-import-resolver-typescript": "^3.6.3",
+    "eslint-import-resolver-typescript": "^3.7.0",
     "eslint-plugin-github": "^4.10.2",
     "eslint-plugin-import": "^2.31.0",
-    "eslint-plugin-prettier": "^5.2.1",
-    "prettier": "^3.3.3",
+    "eslint-plugin-prettier": "^5.2.2",
+    "prettier": "^3.4.2",
     "tsup": "^8.3.5",
-    "typescript": "^5.6.3"
+    "typescript": "^5.7.3"
   }
 }

src/index.ts

@@ -7,6 +7,7 @@ import { SpawnOptions, spawn } from "node:child_process";
 import { mkdirSync, openSync, readFileSync } from "node:fs";
 import * as fs from "node:fs/promises";
 import * as path from "node:path";
+import { setTimeout } from "node:timers/promises";
 
 // The ENV_DAEMON_DIR is intended to determine if we "own" the daemon or not,
 // in the case that a user has put the magic nix cache into their workflow
@@ -14,6 +15,7 @@ import * as path from "node:path";
 const ENV_DAEMON_DIR = "MAGIC_NIX_CACHE_DAEMONDIR";
 
 const FACT_ENV_VARS_PRESENT = "required_env_vars_present";
+const FACT_SENT_SIGTERM = "sent_sigterm";
 const FACT_DIFF_STORE_ENABLED = "diff_store";
 const FACT_ALREADY_RUNNING = "noop_mode";
 
@@ -363,21 +365,25 @@ class MagicNixCacheAction extends DetSysAction {
     actionsCore.debug(`killing daemon process ${pid}`);
 
     try {
+      // Repeatedly signal 0 the daemon to test if it is up.
+      // If it exits, kill will raise an exception which breaks us out of this control flow and skips the sigterm.
+      // If magic-nix-cache doesn't exit in 30s, we SIGTERM it.
+      for (let i = 0; i < 30 * 10; i++) {
+        process.kill(pid, 0);
+        await setTimeout(100);
+      }
+
+      this.addFact(FACT_SENT_SIGTERM, true);
+      actionsCore.info(`Sending Magic Nix Cache a SIGTERM`);
       process.kill(pid, "SIGTERM");
-    } catch (e: unknown) {
-      if (typeof e === "object" && e && "code" in e && e.code !== "ESRCH") {
-        // Throw an error only in strict mode, otherwise ignore because
-        // we're in the post phase and shutting down after this anyway
-        if (this.strictMode) {
-          throw e;
-        }
-      }
-    } finally {
-      if (actionsCore.isDebug()) {
-        actionsCore.info("Entire log:");
-        const entireLog = readFileSync(path.join(this.daemonDir, "daemon.log"));
-        actionsCore.info(entireLog.toString());
-      }
+    } catch {
+      // Perfectly normal to get an exception here, because the process shut down.
+    }
+
+    if (actionsCore.isDebug()) {
+      actionsCore.info("Entire log:");
+      const entireLog = readFileSync(path.join(this.daemonDir, "daemon.log"));
+      actionsCore.info(entireLog.toString());
     }
   }