konduktor-nightly 0.1.0.dev20250915104603__py3-none-any.whl → 0.1.0.dev20251107104752__py3-none-any.whl

konduktor/task.py

@@ -29,7 +29,7 @@ import konduktor
 from konduktor import constants, logging
 from konduktor.data import data_utils
 from konduktor.data import storage as storage_lib
-from konduktor.utils import common_utils, exceptions, schemas, ux_utils, validator
+from konduktor.utils import common_utils, exceptions, schemas, ux_utils
 
 logger = logging.get_logger(__name__)
 
@@ -388,10 +388,6 @@ class Task:
             )
             resources_config['_cluster_config_overrides'] = cluster_config_override
 
-        # Validate Docker image if specified in resources
-        if 'image_id' in resources_config and resources_config['image_id']:
-            validator.validate_and_warn_image(resources_config['image_id'], 'task')
-
         task.set_resources(konduktor.Resources.from_yaml_config(resources_config))
 
         # Parse serving field.
@@ -567,6 +563,13 @@ class Task:
                     f'less than min_replicas ({serving.min_replicas})'
                 )
 
+        if serving.max_replicas == 0 and serving.min_replicas == 0:
+            with ux_utils.print_exception_no_traceback():
+                raise ValueError(
+                    f'max_replicas ({serving.max_replicas}) and '
+                    f'min_replicas ({serving.min_replicas}) cannot both be 0'
+                )
+
         if isinstance(serving, konduktor.Serving):
             serving = serving
         self.serving = serving

konduktor/templates/deployment.yaml.j2

@@ -12,6 +12,10 @@ metadata:
     {{ deployment_num_accelerators_label }}: "{{ num_accelerators }}"
     trainy.ai/has-autoscaler: "{{ autoscaler }}"
     trainy.ai/konduktor-managed: "true"
+    {% if autoscaler == 'true' %}
+    trainy.ai/original-min-replicas: "{{ min_replicas }}"
+    trainy.ai/original-max-replicas: "{{ max_replicas }}"
+    {% endif %}
   name: {{ name }}
   namespace: default
 spec:
@@ -37,11 +41,9 @@ metadata:
     {{ deployment_name_label }}: "{{ name }}"
     {{ deployment_user_label }}: "{{ user }}"
     trainy.ai/has-autoscaler: "{{ autoscaler }}"
-  {% if not general %}
   annotations:
     prometheus.io/scrape: "true"
-    prometheus.io/port: "8080"
-  {% endif %}
+    prometheus.io/port: "9000"
   name: {{ name }}
   namespace: default
 spec:
@@ -61,18 +63,15 @@ spec:
     {{ model_name_label }}: {{ name }}
     {% endif %}
     {{ deployment_name_label }}: "{{ name }}"
-  {% if general %}
-  type: LoadBalancer
-  {% else %}
   type: ClusterIP
-  {% endif %}
 
-{% if not general %}
+# AIBRIX PODAUTOSCALER STUFF (KPA)
+{% if not general and autoscaler == 'true' %}
 ---
 apiVersion: autoscaling.aibrix.ai/v1alpha1
 kind: PodAutoscaler
 metadata:
-  name: {{ name }}-apa
+  name: {{ name }}-pa
   namespace: default
   labels:
     {{ model_name_label }}: {{ name }}
@@ -80,63 +79,113 @@ metadata:
     app.kubernetes.io/managed-by: kustomize
     {{ deployment_name_label }}: "{{ name }}"
     {{ deployment_user_label }}: "{{ user }}"
-  annotations:
-    autoscaling.aibrix.ai/up-fluctuation-tolerance: '0.1'
-    autoscaling.aibrix.ai/down-fluctuation-tolerance: '0.2'
-    apa.autoscaling.aibrix.ai/window: 30s
 spec:
-  scalingStrategy: APA
+  scalingStrategy: KPA
   minReplicas: {{ min_replicas }}
   maxReplicas: {{ max_replicas }}
   metricsSources:
-    - metricSourceType: pod
+    - metricSourceType: domain
       protocolType: http
-      port: "{{ ports }}"
-      path: metrics
-      targetMetric: gpu_cache_usage_perc
-      targetValue: '0.5'
+      endpoint: aibrix-activator.aibrix-activator.svc.cluster.local:8080
+      path: /metrics/default/{{ name }}
+      targetMetric: vllm:deployment_replicas
+      targetValue: "1"
   scaleTargetRef:
     apiVersion: apps/v1
     kind: Deployment
     name: {{ name }}
 {% endif %}
 
+# KEDA HTTP ADD-ON STUFF (1 per deployment)
 {% if general %}
+{% if autoscaler == 'true' %}
+# HTTPScaledObject (1 per deployment) - only when autoscaling enabled
 ---
-apiVersion: autoscaling/v2
-kind: HorizontalPodAutoscaler
+apiVersion: http.keda.sh/v1alpha1
+kind: HTTPScaledObject
 metadata:
-  name: {{ name }}-hpa
+  name: {{ name }}-httpscaledobject
   namespace: default
   labels:
     {{ deployment_name_label }}: "{{ name }}"
     {{ deployment_user_label }}: "{{ user }}"
-    trainy.ai/has-autoscaler: "{{ autoscaler }}"
 spec:
+  hosts:
+    - {{ name }}
+  pathPrefixes:
+    - "/"
+    {% if probe_path %}
+    - "{{ probe_path }}"
+    {% endif %}
   scaleTargetRef:
-    apiVersion: apps/v1
+    name: "{{ name }}"
     kind: Deployment
-    name: {{ name }}
-  minReplicas: {{ min_replicas }}
-  maxReplicas: {{ max_replicas }}
-  metrics:
-    - type: Resource
-      resource:
-        name: cpu
-        target:
-          type: Utilization
-          averageUtilization: 50
-  behavior:
-    scaleDown:
-      stabilizationWindowSeconds: 60
-      policies:
-      - type: Percent
-        value: 100
-        periodSeconds: 15
-    scaleUp:
-      stabilizationWindowSeconds: 20
-      policies:
-      - type: Percent
-        value: 100
-        periodSeconds: 15
+    apiVersion: apps/v1
+    service: "{{ name }}"
+    port: {{ ports }}
+  replicas:
+    min: {{ min_replicas }}
+    max: {{ max_replicas }}
+  scaledownPeriod: 1200 # 20 minutes
+  scalingMetric:
+    requestRate:
+      targetValue: 4
+      granularity: "1s"
+      window: "30s"
+{% endif %}
+
+# INGRESS (1 per deployment)
+---
+apiVersion: networking.k8s.io/v1
+kind: Ingress
+metadata:
+  name: {{ name }}-ingress
+  labels:
+    {{ deployment_name_label }}: "{{ name }}"
+    {{ deployment_user_label }}: "{{ user }}"
+    trainy.ai/konduktor-managed: "true"
+  annotations:
+    nginx.ingress.kubernetes.io/use-regex: "true"
+    nginx.ingress.kubernetes.io/rewrite-target: /$1
+    {% if autoscaler == 'true' %}
+    nginx.ingress.kubernetes.io/upstream-vhost: "{{ name }}"
+    {% endif %}
+spec:
+  ingressClassName: nginx
+  rules:
+  - host: {{ general_base_host }}
+    http:
+      paths:
+      - path: /{{ name }}(.*)
+        pathType: ImplementationSpecific
+        backend:
+          service:
+            {% if autoscaler == 'true' %}
+            # Use KEDA interceptor for autoscaling
+            name: keda-proxy
+            port:
+              number: 8080
+            {% else %}
+            # Direct to app service for fixed replicas
+            name: {{ name }}
+            port:
+              number: {{ ports }}
+            {% endif %}
+  # Direct access convenience rule (via LB IP + Host: {{ name }})
+  - host: {{ name }}
+    http:
+      paths:
+      - path: /(.*)
+        pathType: ImplementationSpecific
+        backend:
+          service:
+            {% if autoscaler == 'true' %}
+            name: keda-proxy
+            port:
+              number: 8080
+            {% else %}
+            name: {{ name }}
+            port:
+              number: {{ ports }}
+            {% endif %}
 {% endif %}

konduktor/templates/pod.yaml.j2

@@ -28,16 +28,21 @@ kubernetes:
       containers:
         # TODO(asaiacai): should decide here whether we add the fabric interfaces/containers init etc.        
         - name: konduktor-container
-          {% if enable_ssh %}
+          {% if enable_ssh or serving %}
           ports:
+            {% if enable_ssh %}
             - name: ssh
               containerPort: {{ konduktor_ssh_port }}
+            {% endif %}
+
+            {% if serving %}
+            - name: serving
+              containerPort: {{ ports }}
+            {% endif %}
           {% endif %}
-          {% if serving %}
-          ports:
-            - containerPort: {{ ports }}
+
+          {% if serving and probe %}
           # TODO (ryan): allow modification of thresholds and timings
-          {% if probe %}
           livenessProbe:
             httpGet:
               path: {{ probe }}
@@ -68,7 +73,6 @@ kubernetes:
             successThreshold: 1
             timeoutSeconds: 1
           {% endif %}
-          {% endif %}
           image: {{ image_id }}
           # this is set during jobset definition since we need to know the jobset
           # name and number of nodes to set all the environment variables correctly here
@@ -77,6 +81,10 @@ kubernetes:
           # flush logs immediately to stdout for more reactive log streaming
           - name: PYTHONUNBUFFERED
             value: "0"
+          - name: KONDUKTOR_NODENAME
+            valueFrom:
+              fieldRef:
+                fieldPath: spec.nodeName
           - name: KONDUKTOR_JOB_NAME
             value: "{{ job_name }}"
           - name: NODE_HOST_IPS
@@ -134,6 +142,8 @@ kubernetes:
           {% if default_secrets %}
           - name: KONDUKTOR_DEFAULT_SECRETS
             value: "/konduktor/default-secrets"
+          - name: KONDUKTOR_DEFAULT_SECRETS_EXPANDED
+            value: "/run/konduktor/expanded-default-secrets"
           {% endif %}
           # these are for compatibility with skypilot
           - name: SKYPILOT_NODE_IPS
@@ -146,6 +156,10 @@ kubernetes:
             value: "{{ num_nodes }}"
           - name: SKYPILOT_NUM_GPUS_PER_NODE
             value: "{{ num_gpus }}"
+          - name: RESTART_ATTEMPT
+            valueFrom:
+              fieldRef:
+                fieldPath: metadata.labels['jobset.sigs.k8s.io/restart-attempt']
           volumeMounts:
           - name: shared-memory
             mountPath: /dev/shm
@@ -159,6 +173,10 @@ kubernetes:
           - name: default-secret-{{ secret.mount_name }}
             mountPath: /konduktor/default-secrets/{{ secret.mount_name }}
           {% endfor %}
+          {% if default_secrets %}
+          - name: default-secrets-expanded
+            mountPath: /run/konduktor/expanded-default-secrets
+          {% endif %}
           {% if git_ssh %}
           - name: git-ssh-secret
             mountPath: /run/konduktor/git-ssh-secret
@@ -192,7 +210,7 @@ kubernetes:
               {% if 'curl' in run_cmd or 'curl' in setup_cmd or tailscale_secret %}
               PACKAGES="$PACKAGES curl";
               {% endif %}
-              {% if 'gs' in mount_secrets or 's3' in mount_secrets %}
+              {% if 'gs' in mount_secrets or 's3' in mount_secrets or default_secrets %}
               PACKAGES="$PACKAGES unzip wget";
               {% endif %}
               {% if 'git' in run_cmd or 'git' in setup_cmd %}
@@ -231,7 +249,7 @@ kubernetes:
               fi;
               end_epoch=$(date +%s);
               
-              echo "Exposing ENV variables"
+              echo "===== KONDUKTOR: Exposing ENV variables ====="
               $(prefix_cmd) env -0 | awk -v RS='\0' '
                 {
                   gsub(/\\/,"\\\\");      # escape existing backslashes first
@@ -346,8 +364,41 @@ kubernetes:
 
               $(prefix_cmd) echo "===== KONDUKTOR: Installing packages took $((end_epoch - start_epoch)) seconds ====="
 
+              $(prefix_cmd) echo "===== KONDUKTOR: Environment variable summary ====="
+              start_epoch=$(date +%s);
+
+              print_bucket () {
+                title="$1"; list="${2:-}"
+                echo "--- $title ---"
+                if [ -n "$list" ]; then
+                  echo "$list" | tr ',' '\n' | sed "s/^/[$title] /"
+                else
+                  echo "[none]"
+                fi
+              }
+
+              # Secrets: prefer detailed mapping if available
+              echo "--- env secret ---"
+              if [ -n "${KONDUKTOR_ENV_SECRETS_MAP_HOPEFULLY_NO_NAME_COLLISION:-}" ]; then
+                echo "${KONDUKTOR_ENV_SECRETS_MAP_HOPEFULLY_NO_NAME_COLLISION}" \
+                  | tr ',' '\n' \
+                  | awk -F'=' '{ printf("[secret: %s] %s\n", $2, $1) }'
+              elif [ -n "${KONDUKTOR_ENV_SECRETS_HOPEFULLY_NO_NAME_COLLISION:-}" ]; then
+                echo "${KONDUKTOR_ENV_SECRETS_HOPEFULLY_NO_NAME_COLLISION}" \
+                  | tr ',' '\n' | sed 's/^/[secret] /'
+              else
+                echo "[none]"
+              fi
+
+              print_bucket "CLI + task.yaml" "${KONDUKTOR_ENV_TASK_ALL_HOPEFULLY_NO_NAME_COLLISION}"
+              print_bucket "config.yaml"     "${KONDUKTOR_ENV_CONFIG_HOPEFULLY_NO_NAME_COLLISION}"
+              print_bucket "other"           "${KONDUKTOR_ENV_OTHER_HOPEFULLY_NO_NAME_COLLISION}"
+
+              end_epoch=$(date +%s);
+              $(prefix_cmd) echo "===== KONDUKTOR: Environment variable summary took $((end_epoch - start_epoch)) seconds ====="
+
               # unpack secrets credentials
-              $(prefix_cmd) echo "===== KONDUKTOR: Unpacking secrets credentials ====="
+              $(prefix_cmd) echo "===== KONDUKTOR: Unpacking cloud storage secret credentials ====="
               start_epoch=$(date +%s);
               mkdir -p ~/.konduktor
               mkdir -p {{ remote_workdir }}
@@ -362,12 +413,71 @@ kubernetes:
               $(prefix_cmd) unzip /run/konduktor/s3-secret/awscredentials -d ~/.aws
               {% endif %}
               {% endfor %}
+              
+              {% if default_secrets %}
+              $(prefix_cmd) echo "===== KONDUKTOR: Unpacking default secrets ====="
+              $(prefix_cmd) mkdir -p "${KONDUKTOR_DEFAULT_SECRETS_EXPANDED}"
+
+              # For each mounted default secret folder:
+              #  - if payload.zip exists, unzip it into the expanded dir
+              #  - otherwise, copy the files as-is
+              for src in "${KONDUKTOR_DEFAULT_SECRETS}"/*; do
+                [ -d "$src" ] || continue
+                name="$(basename "$src")"
+                dst="${KONDUKTOR_DEFAULT_SECRETS_EXPANDED}/${name}"
+                $(prefix_cmd) mkdir -p "$dst"
+
+                if [ -f "${src}/payload.zip" ]; then
+                  $(prefix_cmd) unzip -oq "${src}/payload.zip" -d "$dst"
+                else
+                  $(prefix_cmd) cp -a "${src}/." "$dst/"
+                fi
+              done
+
+              # Point callers to the expanded (writable) path going forward
+              export KONDUKTOR_DEFAULT_SECRETS="${KONDUKTOR_DEFAULT_SECRETS_EXPANDED}"
+              $(prefix_cmd) echo "KONDUKTOR_DEFAULT_SECRETS=${KONDUKTOR_DEFAULT_SECRETS_EXPANDED}" >> /etc/environment
+              {% endif %}
+              
               {% if git_ssh %}
               $(prefix_cmd) echo "Unpacking GIT-SSH secret"
               {% endif %}
               end_epoch=$(date +%s);
               $(prefix_cmd) echo "===== KONDUKTOR: Unpacking secrets credentials took $((end_epoch - start_epoch)) seconds ====="
 
+              $(prefix_cmd) echo "===== KONDUKTOR: Default secret summary ====="
+              start_epoch=$(date +%s)
+
+              root="${KONDUKTOR_DEFAULT_SECRETS:-}"
+              if [[ -z "$root" || ! -d "$root" ]]; then
+                $(prefix_cmd) echo "NO DEFAULT SECRETS FOUND."
+              else
+                for dir in "$root"/*; do
+                  [ -d "$dir" ] || continue
+                  name="$(basename "$dir")"
+
+                  # Pretty header that mirrors the logical mount base:
+                  $(prefix_cmd) echo "/konduktor/default-secrets/${name}:"
+
+                  # Print relative paths only; skip macOS junk and k8s secret internals
+                  (
+                    cd "$dir"
+                    out="$(find . \
+                      \( -name '.DS_Store' -o -name '__MACOSX' -o -name '..data' -o -name '..*' \) -prune -o \
+                      \( -type f -o -type l \) -print \
+                      | sed 's|^\./||' \
+                      | sort)"
+                    if [ -n "$out" ]; then
+                      printf "%s\n" "$out"
+                    fi
+                  )
+                done
+              fi
+
+              end_epoch=$(date +%s)
+              $(prefix_cmd) echo "===== KONDUKTOR: Default secret summary took $((end_epoch - start_epoch)) seconds ====="
+
+
               # sync file mounts
               {% for mkdir_command in mkdir_commands %}
               $(prefix_cmd) {{ mkdir_command }}
@@ -436,6 +546,10 @@ kubernetes:
         secret:
           secretName: {{ secret.k8s_name }}
       {% endfor %}
+      {% if default_secrets %}
+      - name: default-secrets-expanded
+        emptyDir: {}
+      {% endif %}
       {% if git_ssh %}
       - name: git-ssh-secret
         secret:

konduktor/utils/base64_utils.py

@@ -44,6 +44,8 @@ def zip_base64encode(files: List[str]) -> str:
                 else:
                     for root, _, files in os.walk(item_path):
                         for file in files:
+                            if file == '.DS_Store':
+                                continue
                             file_path = os.path.join(root, file)
                             arcname = os.path.relpath(file_path, temp_dir)
                             zipf.write(file_path, arcname)

konduktor/utils/schemas.py

@@ -389,7 +389,7 @@ def get_serving_schema():
         'properties': {
             'min_replicas': {
                 'type': 'integer',
-                'minimum': 1,
+                'minimum': 0,
                 'description': 'Minimum number of replicas for autoscaling.',
             },
             'max_replicas': {

konduktor/utils/validator.py

@@ -37,6 +37,11 @@ SCHEMA_URLS = {
 logger = logging.get_logger(__name__)
 
 
+def _skip_image_checks() -> bool:
+    val = os.getenv('KONDUKTOR_SKIP_IMAGE_CHECK', '')
+    return val.lower() in ('1', 'true', 'yes', 'y')
+
+
 def case_insensitive_enum(validator, enums, instance, schema):
     del validator, schema  # Unused.
     if instance.lower() not in [enum.lower() for enum in enums]:
@@ -419,6 +424,13 @@ def validate_and_warn_image(image_id: str, context: str = 'task') -> None:
     if not image_id:
         return
 
+    if _skip_image_checks():
+        logger.info(
+            'Skipping Docker image validation for %s',
+            image_id,
+        )
+        return
+
     status, message = validate_docker_image(image_id)
 
     if status == 'invalid':

{konduktor_nightly-0.1.0.dev20250915104603.dist-info → konduktor_nightly-0.1.0.dev20251107104752.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: konduktor-nightly
-Version: 0.1.0.dev20250915104603
+Version: 0.1.0.dev20251107104752
 Summary: GPU Cluster Health Management
 Author: Andrew Aikawa
 Author-email: asai@berkeley.edu

{konduktor_nightly-0.1.0.dev20250915104603.dist-info → konduktor_nightly-0.1.0.dev20251107104752.dist-info}/RECORD

@@ -1,4 +1,4 @@
-konduktor/__init__.py,sha256=gW1PYl7NzKJnsnwHEY8-7kKncqEDCGzCOpYT5TGj6So,1574
+konduktor/__init__.py,sha256=mHmTi0owXeaxTt6NwGboUKlwfKWw6xwzbdcUjq9-1DM,1574
 konduktor/adaptors/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 konduktor/adaptors/aws.py,sha256=s47Ra-GaqCQibzVfmD0pmwEWHif1EGO5opMbwkLxTCU,8244
 konduktor/adaptors/common.py,sha256=ZIqzjx77PIHUwpjfAQ1uX8B2aX78YMuGj4Bppd-MdyM,4183
@@ -6,14 +6,14 @@ konduktor/adaptors/gcp.py,sha256=ierTF4z7vwpJ9BsC7LSiwv4uLcjGXscwZOwQrddr2vM,410
 konduktor/authentication.py,sha256=_mVy3eqoKohicHostFiGwG1-2ybxP-l7ouofQ0LRlCY,4570
 konduktor/backends/__init__.py,sha256=usWJ8HdZJEyg7MIsN8Zcz9rk9e2Lq5dWJ8dv6hCN3ys,199
 konduktor/backends/backend.py,sha256=qh0bp94lzoTYZkzyQv2-CVrB5l91FkG2vclXg24UFC0,2910
-konduktor/backends/constants.py,sha256=nt9G9AmFCOMwO4GuKgRQSzJJuKapOmaROp4_Y0tMF5A,732
-konduktor/backends/deployment.py,sha256=fswN9hX_7NwcEogYmo1xn3WgWF8XCcGDvV5yx54_CA0,5860
-konduktor/backends/deployment_utils.py,sha256=qcuoLPeMvEVqgD_h71hQZXAp4ZCdXsFeSBxhtXW6pAA,39846
+konduktor/backends/constants.py,sha256=uAU-edQ_9DNYnu6x7fwNYXIEM7KMfJMOrnG74rlZ8mY,766
+konduktor/backends/deployment.py,sha256=d0a3F7dxDbnRKIt4ZO_kQ0_vet0pZvg4bWYzVZ8DZIQ,6640
+konduktor/backends/deployment_utils.py,sha256=9CmB9CYC_3wxIfIOmTSCN2hbURZ5MpEMTvPwYMUXBRM,49272
 konduktor/backends/jobset.py,sha256=drt8Gc0iYQx18JWXBU6XfhUvC2xCKd8szSJ2JC4O20Q,8640
-konduktor/backends/jobset_utils.py,sha256=If4pv5peB_yXrJJwjkySgVzbjcxDEDWfkOQxUkwSlOk,26386
-konduktor/backends/pod_utils.py,sha256=KP_PAgsdNHFgt4Od-5gAtpifAKIL7DMBg7NJ44uqikg,14885
+konduktor/backends/jobset_utils.py,sha256=g49NY8RFhL_NNd4c1adRLG_Bq3UTFtRURxcAzxnMEYw,26524
+konduktor/backends/pod_utils.py,sha256=kOi3cLbTI3abZFCNQswWrkrOiBBm3gW_9N4INjxeS-w,19276
 konduktor/check.py,sha256=JennyWoaqSKhdyfUldd266KwVXTPJpcYQa4EED4a_BA,7569
-konduktor/cli.py,sha256=OwVdT4ibAQoAJO79YzlThQv_VKlHpsD-CHRwOzehGQ8,57613
+konduktor/cli.py,sha256=B3Pp3RCwkGj8r9YgH-TgC85XU4zcc3eema1kpcDTQ3I,58452
 konduktor/config.py,sha256=9upqgCCYvcu6fKw7tovEYC1MWTkAAir0_WHPdayylbI,15536
 konduktor/constants.py,sha256=T3AeXXxuQHINW_bAWyztvDeS8r4g8kXBGIwIq13cys0,1814
 konduktor/controller/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -68,25 +68,25 @@ konduktor/data/storage_utils.py,sha256=n4GivkN0KMqmyOTDznF0Z-hzsJvm7KCEh5i5HgFAT
 konduktor/execution.py,sha256=d0EP79iSrW2uFsoqn0YV_4kgIupPIqpMOParXx0y3kg,18519
 konduktor/kube_client.py,sha256=HtM3d-_GigHnfGINRANchApR9_OigqczBgeYJ6Dj4j0,8504
 konduktor/logging.py,sha256=xtcCdnecmC3rqMTyunK-klQRINojI7NI4Apag78i9jM,3221
-konduktor/manifests/apoxy-setup.yaml,sha256=HDZu7Evm_siIpK1E4tNZ9WVTnFV2LhBXwjJlFOYSTcU,3319
-konduktor/manifests/apoxy-setup2.yaml,sha256=BhXsgcVrLBruLXnF7xlj0Ej6YVJFYMABJIpYtwakQMo,731
+konduktor/manifests/aibrix-setup.yaml,sha256=Foe3M1C0zVt-CVSJGr5SlQkMaNGs9kU2CvIZoANm3f8,14133
+konduktor/manifests/apoxy-setup.yaml,sha256=EipknCq33aBdxu9BIo6y5novjO0B_d_DCWqY44zYNuU,4262
+konduktor/manifests/apoxy-setup2.yaml,sha256=fc1tDwVopPVFzvUygkqxDGVqHHeo1cF9ERTnzUdgaGs,2517
 konduktor/manifests/controller_deployment.yaml,sha256=6p3oSLkEVONZsvKZGqVop0Dhn4bo3lrigRmhf8NXBHE,1730
 konduktor/manifests/dashboard_deployment.yaml,sha256=xJLd4FbPMAosI0fIv5_8y7dV9bw0Vsf81l-w4MB_aU8,2837
 konduktor/manifests/dmesg_daemonset.yaml,sha256=pSWt7YOeTYjS0l0iki1fvHOs7MhY-sH-RQfVW6JJyno,1391
 konduktor/manifests/pod_cleanup_controller.yaml,sha256=hziL1Ka1kCAEL9R7Tjvpb80iw1vcq9_3gwHCu75Bi0A,3939
-konduktor/resource.py,sha256=kfdhnUR_9kDLSm2sUAkv1sLQXyAkI08p3wIzCz1p7-M,20791
-konduktor/serving.py,sha256=sh8TPAUXg23Bkt0ByatIMdxFFqzRm18HJTEkt3wHzdo,5147
-konduktor/task.py,sha256=oFRHdMevg7lGYkHugCHl89FUREfq9M-l0Qd3N-rjHMA,37727
-konduktor/templates/apoxy-deployment.yaml.j2,sha256=_EdT7w0rBK3if1INHT6GGUEugy0mOkRfYOWRgBcKLdo,942
-konduktor/templates/deployment.yaml.j2,sha256=uXFjDQaimbpFdAn2RJGaIvS_PzDY136cw_L3QMjz3ZA,3452
+konduktor/resource.py,sha256=JqEE3LZiBBd5vqAiHDk-nlLve_VUQHhgdo6BIgx2Xfk,21215
+konduktor/serving.py,sha256=4s8cQhsVjf-HByZF65pbMxuqaV319hUSQE9pC8gP4Sg,5405
+konduktor/task.py,sha256=FIWm_rC_63GPBoe-Hi8a_eJ0H8Szw747SwXYPrNtOWE,37820
+konduktor/templates/deployment.yaml.j2,sha256=0Cer53I8YHtYgUeEBQ_NVgC36FdOcjMNejgfP8teJC4,4964
 konduktor/templates/jobset.yaml.j2,sha256=NQcVeRNsTLLmTnJRnkL1vr45mSeth-b11YShXn_RoSg,1323
-konduktor/templates/pod.yaml.j2,sha256=p9yE-AQkCF2Tgjd1QQiOLtzgI6Gbpps-MZYJZ9fQWIs,19159
+konduktor/templates/pod.yaml.j2,sha256=gGYwdXsPxStiua9Mm-OF7byVfuKjcH-TYNjoQmdpX_Q,24107
 konduktor/usage/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 konduktor/usage/constants.py,sha256=gCL8afIHZhO0dcxbJGpESE9sCC1cBSbeRnQ8GwNOY4M,612
 konduktor/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 konduktor/utils/accelerator_registry.py,sha256=ythz3ynulP1DSSU7Jj5VUsQeBzSYRkxCVDZ5oOg0xtc,560
 konduktor/utils/annotations.py,sha256=oy2-BLydkFt3KWkXDuaGY84d6b7iISuy4eAT9uXk0Fc,2225
-konduktor/utils/base64_utils.py,sha256=mF-Tw98mFRG70YE4w6s9feuQSCYZHOb8YatBZwMugyI,3130
+konduktor/utils/base64_utils.py,sha256=TzKxe_SPHrurJ1lTOwOi4OmGUNkGxomstFcTWcWLQhw,3223
 konduktor/utils/common_utils.py,sha256=8gBpzYiC1bQ8sbgHIFLkKCGT5nLs1afpejod60kVSos,15076
 konduktor/utils/constants.py,sha256=1DneiTR21lvKUcWdBGwC4I4fD4uPjbjLUilEnJS7rzA,216
 konduktor/utils/env_options.py,sha256=T41Slzf4Mzl-n45CGXXqdy2fCrYhPNZQ7RP5vmnN4xc,2258
@@ -96,12 +96,12 @@ konduktor/utils/kubernetes_utils.py,sha256=XleYxzG64hciZb-CjzBDjX8BOMhFATIIHZlXD
 konduktor/utils/log_utils.py,sha256=VUyTtN819BJnSwm33-73-h8aaD51Y5Gawt6ek2kU1tk,18181
 konduktor/utils/loki_utils.py,sha256=eOGiD7dZNuwzmyXKiifyqz00EVh2nwcUPFSiPkac9y0,4050
 konduktor/utils/rich_utils.py,sha256=ycADW6Ij3wX3uT8ou7T8qxX519RxlkJivsLvUahQaJo,3583
-konduktor/utils/schemas.py,sha256=X2q-Nuk71EfMQXl4QFOtFWlQgd8tC_jFAo5dScmlEQc,19067
+konduktor/utils/schemas.py,sha256=cr39nEAgjluhXoUYnvIwCwLBH8rLds37MBsF1uQv1rw,19067
 konduktor/utils/subprocess_utils.py,sha256=WoFkoFhGecPR8-rF8WJxbIe-YtV94LXz9UG64SDhCY4,9448
 konduktor/utils/ux_utils.py,sha256=LSH4b5lckD157qDF4keThxtkGdxNrAfGKmH1ewhZkm4,8646
-konduktor/utils/validator.py,sha256=gCB5v9Up9bCWD_92fS5ChfRRXj_m56Ky9uzd_77wXGI,16927
-konduktor_nightly-0.1.0.dev20250915104603.dist-info/LICENSE,sha256=MuuqTZbHvmqXR_aNKAXzggdV45ANd3wQ5YI7tnpZhm0,6586
-konduktor_nightly-0.1.0.dev20250915104603.dist-info/METADATA,sha256=noibgctWCSonSUXnlV-7BQS_PAMNwkajGxAHNpvdwWM,4247
-konduktor_nightly-0.1.0.dev20250915104603.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
-konduktor_nightly-0.1.0.dev20250915104603.dist-info/entry_points.txt,sha256=k3nG5wDFIJhNqsZWrHk4d0irIB2Ns9s47cjRWYsTCT8,48
-konduktor_nightly-0.1.0.dev20250915104603.dist-info/RECORD,,
+konduktor/utils/validator.py,sha256=UcLvZCk9Cpbbhw8r_ZJtTpMSTfY1NKqcyciKsPzRPZM,17222
+konduktor_nightly-0.1.0.dev20251107104752.dist-info/LICENSE,sha256=MuuqTZbHvmqXR_aNKAXzggdV45ANd3wQ5YI7tnpZhm0,6586
+konduktor_nightly-0.1.0.dev20251107104752.dist-info/METADATA,sha256=EEA9KjVBKhzBk4hO1-mWEacCmBul0d5GqMbB_VUKWbQ,4247
+konduktor_nightly-0.1.0.dev20251107104752.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
+konduktor_nightly-0.1.0.dev20251107104752.dist-info/entry_points.txt,sha256=k3nG5wDFIJhNqsZWrHk4d0irIB2Ns9s47cjRWYsTCT8,48
+konduktor_nightly-0.1.0.dev20251107104752.dist-info/RECORD,,

konduktor/templates/apoxy-deployment.yaml.j2

@@ -1,33 +0,0 @@
----
-# Apoxy Backend for general deployment
-apiVersion: core.apoxy.dev/v1alpha
-kind: Backend
-metadata:
-  name: {{ unique_cluster_name }}-backend-{{ deployment_number }}
-  labels:
-    task_name: {{ name }}
-    endpoint_name: {{ cluster_name }}-{{ deployment_number }}.trainy.us
-spec:
-  endpoints:
-    - fqdn: {{ name }}.default.{{ unique_cluster_name }}.tun.apoxy.net
----
-# Apoxy Route for general deployment
-apiVersion: gateway.apoxy.dev/v1
-kind: HTTPRoute
-metadata:
-  name: {{ unique_cluster_name }}-route-{{ deployment_number }}
-  labels:
-    task_name: {{ name }}
-    endpoint_name: {{ cluster_name }}-{{ deployment_number }}.trainy.us
-spec:
-  parentRefs:
-    - name: default 
-      kind: Gateway
-      port: 443
-  hostnames:
-    - '{{ cluster_name }}-{{ deployment_number }}.trainy.us'
-  rules:
-    - backendRefs:
-        - kind: Backend
-          name: {{ unique_cluster_name }}-backend-{{ deployment_number }}
-          port: {{ ports }}