chore(infra,web): cap container logs and trim noisy server logs

apps/web/src/actions/opencode.ts

@@ -826,13 +826,6 @@ export async function sendMessageAction(
   message?: WorkspaceMessage;
   error?: string;
 }> {
-  console.log("[sendMessageAction] Called with:", {
-    slug,
-    sessionId,
-    text: text.substring(0, 50),
-    model,
-  });
-
   // Verify user is authorized
   const session = await getSession();
   if (!session) {
@@ -851,11 +844,6 @@ export async function sendMessageAction(
       !instance.serverPassword ||
       instance.status !== "running"
     ) {
-      console.log("[sendMessageAction] Instance unavailable:", {
-        hasInstance: !!instance,
-        hasPassword: !!instance?.serverPassword,
-        status: instance?.status,
-      });
       return { ok: false, error: "instance_unavailable" };
     }
 
@@ -865,11 +853,6 @@ export async function sendMessageAction(
     )}`;
     const baseUrl = getInstanceUrl(slug);
 
-    console.log(
-      "[sendMessageAction] Sending to:",
-      `${baseUrl}/session/${sessionId}/message`
-    );
-
     const body = {
       parts: [{ type: "text", text }],
       model: model
@@ -912,14 +895,8 @@ export async function sendMessageAction(
           }
         }
       }
-
-      console.log(
-        "[sendMessageAction] Extracted text:",
-        textContent.substring(0, 100)
-      );
     } catch {
-      // If not valid JSON, maybe it's streaming format (NDJSON)
-      console.log("[sendMessageAction] JSON parse failed, trying NDJSON");
+      // If not valid JSON, fall back to NDJSON parsing.
       const lines = responseText.split("\n");
 
       for (const line of lines) {

apps/web/src/actions/spawner.ts

@@ -85,29 +85,22 @@ export async function ensureInstanceRunningAction(slug: string): Promise<{
   status: 'running' | 'starting' | 'error'
   error?: string
 }> {
-  console.log('[ensureInstanceRunning] Starting for slug:', slug)
-
   try {
     const session = await getSession()
     if (!session) {
-      console.log('[ensureInstanceRunning] No session - unauthorized')
       return { status: 'error', error: 'unauthorized' }
     }
-    console.log('[ensureInstanceRunning] Session user:', session.user.slug, 'role:', session.user.role)
 
     if (session.user.slug !== slug && session.user.role !== 'ADMIN') {
-      console.log('[ensureInstanceRunning] Forbidden - slug mismatch')
       return { status: 'error', error: 'forbidden' }
     }
 
     const kickstartStatus = await getKickstartStatus()
     if (kickstartStatus !== 'ready') {
-      console.log('[ensureInstanceRunning] Kickstart setup required')
       return { status: 'error', error: 'setup_required' }
     }
 
     const instance = await getWorkspaceStatus(slug)
-    console.log('[ensureInstanceRunning] Current instance status:', instance?.status ?? 'none')
 
     if (instance?.status === 'running') {
       const syncUserId =
@@ -146,9 +139,7 @@ export async function ensureInstanceRunningAction(slug: string): Promise<{
       return { status: 'starting' }
     }
 
-    console.log('[ensureInstanceRunning] Starting instance...')
     const result = await startWorkspace(slug, session.user.id)
-    console.log('[ensureInstanceRunning] Start result:', result)
 
     if (!result.ok) {
       return { status: 'error', error: result.detail ?? result.error }

apps/web/src/lib/opencode/transform.ts

@@ -7,6 +7,11 @@ import { WORKSPACE_ATTACHMENTS_DIR } from "@/lib/workspace-attachments";
  */
 const HIDDEN_PART_TYPES = new Set(["snapshot", "compaction"]);
 
+// Track which unknown part types we have already logged so each new kind only
+// fires once per process. Without this, a single unrecognized part type would
+// log the full object on every streamed message and rapidly fill disk.
+const loggedUnknownPartTypes = new Set<string>();
+
 function resolveFilePartPath(sourcePath: string | undefined, fileUrl: string | undefined): string | undefined {
   if (sourcePath) {
     return sourcePath;
@@ -238,8 +243,13 @@ export function transformParts(parts: unknown[]): MessagePart[] {
         }
 
         default: {
-          // Unknown type - preserve as fallback for debugging
-          console.log("[transformParts] Unknown part type:", partType, part);
+          // Unknown type - preserve as fallback for debugging.
+          // Log only once per kind to avoid flooding logs during streaming.
+          const kind = typeof partType === "string" ? partType : "non-string";
+          if (!loggedUnknownPartTypes.has(kind)) {
+            loggedUnknownPartTypes.add(kind);
+            console.warn("[transformParts] Unknown part type:", kind);
+          }
           const normalizedData = normalizeSerializableValue(part);
           return {
             type: "unknown" as const,

apps/web/src/lib/spawner/__tests__/docker.test.ts

@@ -156,6 +156,13 @@ describe('docker', () => {
             '/opt/arche/kb-content:/kb-content',
             '/opt/arche/users/user-slug:/tmp/arche-user-data:ro',
           ],
+          LogConfig: {
+            Type: 'json-file',
+            Config: {
+              'max-size': '10m',
+              'max-file': '3',
+            },
+          },
         },
         Labels: {
           'arche.managed': 'true',

apps/web/src/lib/spawner/core.ts

@@ -283,20 +283,12 @@ async function waitForHealthy(containerId: string, slug: string, password: strin
 
     if (directBaseUrl === undefined) {
       directBaseUrl = await getContainerHealthBaseUrl(containerId)
-      if (directBaseUrl) {
-        console.log('[spawner] Using direct container IP for initial healthcheck', {
-          baseUrl: directBaseUrl,
-          containerId,
-          slug,
-        })
-      }
     }
 
     if (directBaseUrl && !directHealthy) {
       const directHealth = await isInstanceHealthyWithPassword(slug, password, directBaseUrl)
       if (directHealth.ok) {
         directHealthy = true
-        console.log('[spawner] OpenCode responded on direct container IP', { containerId, slug })
       } else {
         lastHealth = directHealth
       }

apps/web/src/lib/spawner/docker.ts

@@ -213,6 +213,16 @@ export async function createContainer(
       NetworkMode: getOpencodeNetwork(),
       RestartPolicy: { Name: "on-failure", MaximumRetryCount: 5 },
       Binds: binds,
+      // Cap per-workspace container logs to avoid filling host disk.
+      // Podman aliases json-file to k8s-file; max-size is honored, max-file
+      // is best-effort (some Podman versions truncate instead of rotating).
+      LogConfig: {
+        Type: "json-file",
+        Config: {
+          "max-size": "10m",
+          "max-file": "3",
+        },
+      },
     },
     Labels: {
       "arche.managed": "true",

infra/deploy/ansible/roles/app/tasks/main.yml

@@ -391,6 +391,8 @@
       --label arche.role=web \
       --label arche.version={{ web_version }} \
       --restart on-failure:5 \
+      --log-driver k8s-file \
+      --log-opt max-size=10m \
       {{ web_image }}
   changed_when: true
   no_log: true

infra/deploy/ansible/roles/app/templates/compose.yml.j2

@@ -5,10 +5,17 @@
 
 name: arche
 
+x-logging: &default-logging
+  driver: "json-file"
+  options:
+    max-size: "10m"
+    max-file: "3"
+
 services:
   traefik:
     image: docker.io/traefik:v3.6.7
     restart: "on-failure:5"
+    logging: *default-logging
     command:
       - --providers.docker=true
       - --providers.docker.exposedbydefault=false
@@ -58,6 +65,7 @@ services:
   cloudflared:
     image: {{ cloudflared_image }}
     restart: "on-failure:5"
+    logging: *default-logging
     command:
       - tunnel
       - --no-autoupdate
@@ -73,6 +81,7 @@ services:
   docker-socket-proxy:
     image: ghcr.io/tecnativa/docker-socket-proxy:latest
     restart: "on-failure:5"
+    logging: *default-logging
 {% if deploy_mode == 'local-dev' %}
     user: root
     security_opt:
@@ -98,6 +107,7 @@ services:
   postgres:
     image: docker.io/postgres:16
     restart: "on-failure:5"
+    logging: *default-logging
     environment:
       POSTGRES_DB: arche
       POSTGRES_USER: postgres
@@ -142,6 +152,7 @@ services:
   reaper:
     image: {{ web_image }}
     restart: "on-failure:5"
+    logging: *default-logging
     command:
       - pnpm
       - run
@@ -164,6 +175,7 @@ services:
   flows:
     image: {{ web_image }}
     restart: "unless-stopped"
+    logging: *default-logging
     command:
       - pnpm
       - run
@@ -199,6 +211,7 @@ services:
     image: node:24.14.0-bookworm-slim
     working_dir: /app
     restart: "unless-stopped"
+    logging: *default-logging
     env_file:
       - {{ env_file_name | default('.env') }}
     ports: