From e1a1f447246a8fd24be1a70214ab3c57774b2b46 Mon Sep 17 00:00:00 2001
From: sajdakabir <sajda.kbir@gmail.com>
Date: Wed, 6 May 2026 22:54:46 +0530
Subject: [PATCH] fix(deploy): notify on container crash loop after deployment

---
 packages/server/src/services/application.ts | 22 +++++-
 packages/server/src/utils/docker/utils.ts   | 83 +++++++++++++++++++++
 2 files changed, 104 insertions(+), 1 deletion(-)

diff --git a/packages/server/src/services/application.ts b/packages/server/src/services/application.ts
index ac1fbb4492..c6a5ef6059 100644
--- a/packages/server/src/services/application.ts
+++ b/packages/server/src/services/application.ts
@@ -30,7 +30,7 @@ import { createTraefikConfig } from "@dokploy/server/utils/traefik/application";
 import { TRPCError } from "@trpc/server";
 import { eq } from "drizzle-orm";
 import type { z } from "zod";
-import { encodeBase64 } from "../utils/docker/utils";
+import { encodeBase64, waitForSwarmServiceStable } from "../utils/docker/utils";
 import { getDokployUrl } from "./admin";
 import {
 	createDeployment,
@@ -222,6 +222,16 @@ export const deployApplication = async ({
 		}
 
 		await mechanizeDockerContainer(application);
+
+		const stability = await waitForSwarmServiceStable(application.appName, {
+			serverId,
+		});
+		if (!stability.stable) {
+			throw new Error(
+				`Container did not stay running after deployment: ${stability.reason}`,
+			);
+		}
+
 		await updateDeploymentStatus(deployment.deploymentId, "done");
 		await updateApplicationStatus(applicationId, "done");
 
@@ -313,6 +323,16 @@ export const rebuildApplication = async ({
 			await execAsync(commandWithLog);
 		}
 		await mechanizeDockerContainer(application);
+
+		const stability = await waitForSwarmServiceStable(application.appName, {
+			serverId,
+		});
+		if (!stability.stable) {
+			throw new Error(
+				`Container did not stay running after rebuild: ${stability.reason}`,
+			);
+		}
+
 		await updateDeploymentStatus(deployment.deploymentId, "done");
 		await updateApplicationStatus(applicationId, "done");
 
diff --git a/packages/server/src/utils/docker/utils.ts b/packages/server/src/utils/docker/utils.ts
index 8065b7dd93..c23b3e8213 100644
--- a/packages/server/src/utils/docker/utils.ts
+++ b/packages/server/src/utils/docker/utils.ts
@@ -854,6 +854,89 @@ const getSwarmServiceContainerId = async (
 	}
 };
 
+export type SwarmStabilityResult =
+	| { stable: true }
+	| { stable: false; reason: string };
+
+export const waitForSwarmServiceStable = async (
+	appName: string,
+	{
+		serverId,
+		windowMs = 60_000,
+		pollMs = 5_000,
+	}: { serverId?: string | null; windowMs?: number; pollMs?: number } = {},
+): Promise<SwarmStabilityResult> => {
+	const remoteDocker = await getRemoteDocker(serverId);
+	const deadline = Date.now() + windowMs;
+	let everRunning = false;
+	let lastReason = "Service did not reach running state";
+
+	while (Date.now() < deadline) {
+		try {
+			const tasks = await remoteDocker.listTasks({
+				filters: JSON.stringify({ service: [appName] }),
+			});
+
+			const sorted = [...tasks].sort((a, b) => {
+				const at = new Date(a.UpdatedAt ?? 0).getTime();
+				const bt = new Date(b.UpdatedAt ?? 0).getTime();
+				return bt - at;
+			});
+			const latest = sorted[0];
+			const state = latest?.Status?.State;
+			const message = latest?.Status?.Err || latest?.Status?.Message || "";
+
+			if (state === "failed" || state === "rejected") {
+				return {
+					stable: false,
+					reason: message
+						? `Task ${state}: ${message}`
+						: `Task entered ${state} state`,
+				};
+			}
+
+			const runningCount = sorted.filter(
+				(t) => t.Status?.State === "running",
+			).length;
+			const startingCount = sorted.filter((t) =>
+				[
+					"new",
+					"pending",
+					"assigned",
+					"accepted",
+					"preparing",
+					"starting",
+				].includes(t.Status?.State ?? ""),
+			).length;
+
+			if (runningCount > 0) {
+				everRunning = true;
+			} else if (everRunning && startingCount > 0) {
+				return {
+					stable: false,
+					reason: message
+						? `Container restarted after running: ${message}`
+						: "Container restarted after reaching running state",
+				};
+			}
+
+			lastReason = message
+				? `Latest task state: ${state ?? "unknown"} (${message})`
+				: `Latest task state: ${state ?? "unknown"}`;
+		} catch (error) {
+			lastReason =
+				error instanceof Error ? error.message : "Failed to inspect service";
+		}
+
+		await new Promise((resolve) => setTimeout(resolve, pollMs));
+	}
+
+	if (everRunning) {
+		return { stable: true };
+	}
+	return { stable: false, reason: lastReason };
+};
+
 export const checkPostgresHealth = async (): Promise<ServiceHealthStatus> => {
 	const serviceCheck = await checkSwarmServiceRunning("dokploy-postgres");
 	if (serviceCheck.status === "unhealthy") {