prevent task monitor from processing errors handled elsewhere

Author: nicktrn

apps/kubernetes-provider/src/taskMonitor.ts

@@ -140,7 +140,7 @@ export class TaskMonitor {
     const containerState = this.#getContainerStateSummary(containerStatus.state);
     const exitCode = containerState.exitCode ?? -1;
 
-    if (exitCode === EXIT_CODE_ALREADY_HANDLED) {
+    if (exitCode === EXIT_CODE_ALREADY_HANDLED || exitCode === EXIT_CODE_CHILD_NONZERO) {
       this.#logger.debug("Ignoring pod failure, already handled by worker", {
         podName,
       });
@@ -186,9 +186,8 @@ export class TaskMonitor {
         break;
       case "OOMKilled":
         overrideCompletion = true;
-        reason = `${
-          exitCode === EXIT_CODE_CHILD_NONZERO ? "Child process" : "Parent process"
-        } ran out of memory! Try choosing a machine preset with more memory for this task.`;
+        reason =
+          "[TaskMonitor] Your task ran out of memory. Try increasing the machine specs. If this doesn't fix it there might be a memory leak.";
         errorCode = TaskRunErrorCodes.TASK_PROCESS_OOM_KILLED;
         break;
       default:

packages/cli-v3/src/entryPoints/deploy-run-controller.ts

@@ -519,12 +519,12 @@ class ProdWorker {
 
     logger.log("completion acknowledged", { willCheckpointAndRestore, shouldExit });
 
-    const exitCode =
+    const isNonZeroExitError =
       !completion.ok &&
       completion.error.type === "INTERNAL_ERROR" &&
-      completion.error.code === TaskRunErrorCodes.TASK_PROCESS_EXITED_WITH_NON_ZERO_CODE
-        ? EXIT_CODE_CHILD_NONZERO
-        : 0;
+      completion.error.code === TaskRunErrorCodes.TASK_PROCESS_EXITED_WITH_NON_ZERO_CODE;
+
+    const exitCode = isNonZeroExitError ? EXIT_CODE_CHILD_NONZERO : 0;
 
     if (shouldExit) {
       // Exit after completion, without any retrying

packages/core/src/v3/apps/process.ts

@@ -1,2 +1,4 @@
+/** This was used by the old build system in case of indexing failures */
 export const EXIT_CODE_ALREADY_HANDLED = 111;
-export const EXIT_CODE_CHILD_NONZERO = 112;
+/** This means what it says and is only set once we have completed the attempt */
+export const EXIT_CODE_CHILD_NONZERO = 112;