fix: no default resource limits (#768)

README.md

@@ -220,7 +220,7 @@ participants:
     # Resource management for el containers
     # CPU is milicores
     # RAM is in MB
-    # Defaults are set per client
+    # Defaults to 0, which results in no resource limits
     el_min_cpu: 0
     el_max_cpu: 0
     el_min_mem: 0
@@ -278,7 +278,7 @@ participants:
     # Resource management for cl containers
     # CPU is milicores
     # RAM is in MB
-    # Defaults are set per client
+    # Defaults to 0, which results in no resource limits
     cl_min_cpu: 0
     cl_max_cpu: 0
     cl_min_mem: 0
@@ -340,7 +340,7 @@ participants:
     # Resource management for vc containers
     # CPU is milicores
     # RAM is in MB
-    # Defaults are set per client
+    # Defaults to 0, which results in no resource limits
     vc_min_cpu: 0
     vc_max_cpu: 0
     vc_min_mem: 0

src/cl/cl_launcher.star

@@ -95,7 +95,7 @@ def launch(
         or constants.NETWORK_NAME.shadowfork in network_params.network
         else None
     )
-
+    network_name = shared_utils.get_network_name(network_params.network)
     for index, participant in enumerate(participants):
         cl_type = participant.cl_type
         el_type = participant.el_type
@@ -104,6 +104,26 @@ def launch(
             global_node_selectors,
         )
 
+        tolerations = input_parser.get_client_tolerations(
+            participant.cl_tolerations, participant.tolerations, global_tolerations
+        )
+
+        (
+            cl_min_cpu,
+            cl_max_cpu,
+            cl_min_mem,
+            cl_max_mem,
+            cl_volume_size,
+        ) = shared_utils.get_cpu_mem_resource_limits(
+            participant.cl_min_cpu,
+            participant.cl_max_cpu,
+            participant.cl_min_mem,
+            participant.cl_max_mem,
+            participant.cl_volume_size,
+            network_name,
+            participant.cl_type,
+        )
+
         if cl_type not in cl_launchers:
             fail(
                 "Unsupported launcher '{0}', need one of '{1}'".format(
@@ -158,10 +178,10 @@ def launch(
                 el_context,
                 full_name,
                 new_cl_node_validator_keystores,
-                participant.cl_min_cpu,
-                participant.cl_max_cpu,
-                participant.cl_min_mem,
-                participant.cl_max_mem,
+                cl_min_cpu,
+                cl_max_cpu,
+                cl_min_mem,
+                cl_max_mem,
                 participant.snooper_enabled,
                 snooper_engine_context,
                 participant.blobber_enabled,
@@ -170,10 +190,8 @@ def launch(
                 participant.cl_extra_env_vars,
                 participant.cl_extra_labels,
                 persistent,
-                participant.cl_volume_size,
-                participant.cl_tolerations,
-                participant.tolerations,
-                global_tolerations,
+                cl_volume_size,
+                tolerations,
                 node_selectors,
                 participant.use_separate_vc,
                 participant.keymanager_enabled,
@@ -195,10 +213,10 @@ def launch(
                 el_context,
                 full_name,
                 new_cl_node_validator_keystores,
-                participant.cl_min_cpu,
-                participant.cl_max_cpu,
-                participant.cl_min_mem,
-                participant.cl_max_mem,
+                cl_min_cpu,
+                cl_max_cpu,
+                cl_min_mem,
+                cl_max_mem,
                 participant.snooper_enabled,
                 snooper_engine_context,
                 participant.blobber_enabled,
@@ -207,10 +225,8 @@ def launch(
                 participant.cl_extra_env_vars,
                 participant.cl_extra_labels,
                 persistent,
-                participant.cl_volume_size,
-                participant.cl_tolerations,
-                participant.tolerations,
-                global_tolerations,
+                cl_volume_size,
+                tolerations,
                 node_selectors,
                 participant.use_separate_vc,
                 participant.keymanager_enabled,

src/cl/grandine/grandine_launcher.star

@@ -15,10 +15,6 @@ BEACON_DISCOVERY_PORT_NUM = 9000
 BEACON_HTTP_PORT_NUM = 4000
 BEACON_METRICS_PORT_NUM = 8008
 
-# The min/max CPU/memory that the beacon node can use
-BEACON_MIN_CPU = 50
-BEACON_MIN_MEMORY = 1024
-
 BEACON_METRICS_PATH = "/metrics"
 
 MIN_PEERS = 1
@@ -58,9 +54,7 @@ def launch(
     extra_labels,
     persistent,
     cl_volume_size,
-    cl_tolerations,
-    participant_tolerations,
-    global_tolerations,
+    tolerations,
     node_selectors,
     use_separate_vc,
     keymanager_enabled,
@@ -74,33 +68,8 @@ def launch(
         participant_log_level, global_log_level, VERBOSITY_LEVELS
     )
 
-    tolerations = input_parser.get_client_tolerations(
-        cl_tolerations, participant_tolerations, global_tolerations
-    )
-
     extra_params = [param for param in extra_params]
 
-    network_name = shared_utils.get_network_name(launcher.network)
-
-    cl_min_cpu = int(cl_min_cpu) if int(cl_min_cpu) > 0 else BEACON_MIN_CPU
-    cl_max_cpu = (
-        int(cl_max_cpu)
-        if int(cl_max_cpu) > 0
-        else constants.RAM_CPU_OVERRIDES[network_name]["grandine_max_cpu"]
-    )
-    cl_min_mem = int(cl_min_mem) if int(cl_min_mem) > 0 else BEACON_MIN_MEMORY
-    cl_max_mem = (
-        int(cl_max_mem)
-        if int(cl_max_mem) > 0
-        else constants.RAM_CPU_OVERRIDES[network_name]["grandine_max_mem"]
-    )
-
-    cl_volume_size = (
-        int(cl_volume_size)
-        if int(cl_volume_size) > 0
-        else constants.VOLUME_SIZE[network_name]["grandine_volume_size"]
-    )
-
     config = get_beacon_config(
         plan,
         launcher.el_cl_genesis_data,
@@ -386,33 +355,42 @@ def get_beacon_config(
             persistent_key="data-{0}".format(service_name),
             size=cl_volume_size,
         )
-
-    return ServiceConfig(
-        image=image,
-        ports=used_ports,
-        public_ports=public_ports,
-        cmd=cmd,
-        env_vars=extra_env_vars,
-        files=files,
-        private_ip_address_placeholder=constants.PRIVATE_IP_ADDRESS_PLACEHOLDER,
-        ready_conditions=cl_node_ready_conditions.get_ready_conditions(
+    config_args = {
+        "image": image,
+        "ports": used_ports,
+        "public_ports": public_ports,
+        "cmd": cmd,
+        "files": files,
+        "env_vars": extra_env_vars,
+        "private_ip_address_placeholder": constants.PRIVATE_IP_ADDRESS_PLACEHOLDER,
+        "ready_conditions": cl_node_ready_conditions.get_ready_conditions(
             constants.HTTP_PORT_ID
         ),
-        min_cpu=cl_min_cpu,
-        max_cpu=cl_max_cpu,
-        min_memory=cl_min_mem,
-        max_memory=cl_max_mem,
-        labels=shared_utils.label_maker(
+        "labels": shared_utils.label_maker(
             constants.CL_TYPE.grandine,
             constants.CLIENT_TYPES.cl,
             image,
             el_context.client_name,
             extra_labels,
         ),
-        user=User(uid=0, gid=0),
-        tolerations=tolerations,
-        node_selectors=node_selectors,
-    )
+        "tolerations": tolerations,
+        "node_selectors": node_selectors,
+        "user": User(uid=0, gid=0),
+    }
+
+    if cl_min_cpu > 0:
+        config_args["min_cpu"] = cl_min_cpu
+
+    if cl_max_cpu > 0:
+        config_args["max_cpu"] = cl_max_cpu
+
+    if cl_min_mem > 0:
+        config_args["min_memory"] = cl_min_mem
+
+    if cl_max_mem > 0:
+        config_args["max_memory"] = cl_max_mem
+
+    return ServiceConfig(**config_args)
 
 
 def new_grandine_launcher(

src/cl/lighthouse/lighthouse_launcher.star

@@ -60,9 +60,7 @@ def launch(
     extra_labels,
     persistent,
     cl_volume_size,
-    cl_tolerations,
-    participant_tolerations,
-    global_tolerations,
+    tolerations,
     node_selectors,
     use_separate_vc,
     keymanager_enabled,
@@ -77,31 +75,6 @@ def launch(
         participant_log_level, global_log_level, VERBOSITY_LEVELS
     )
 
-    tolerations = input_parser.get_client_tolerations(
-        cl_tolerations, participant_tolerations, global_tolerations
-    )
-
-    network_name = shared_utils.get_network_name(launcher.network)
-
-    cl_min_cpu = int(cl_min_cpu) if int(cl_min_cpu) > 0 else BEACON_MIN_CPU
-    cl_max_cpu = (
-        int(cl_max_cpu)
-        if int(cl_max_cpu) > 0
-        else constants.RAM_CPU_OVERRIDES[network_name]["lighthouse_max_cpu"]
-    )
-    cl_min_mem = int(cl_min_mem) if int(cl_min_mem) > 0 else BEACON_MIN_MEMORY
-    cl_max_mem = (
-        int(cl_max_mem)
-        if int(cl_max_mem) > 0
-        else constants.RAM_CPU_OVERRIDES[network_name]["lighthouse_max_mem"]
-    )
-
-    cl_volume_size = (
-        int(cl_volume_size)
-        if int(cl_volume_size) > 0
-        else constants.VOLUME_SIZE[network_name]["lighthouse_volume_size"]
-    )
-
     # Launch Beacon node
     beacon_config = get_beacon_config(
         plan,
@@ -374,31 +347,41 @@ def get_beacon_config(
         )
     env = {RUST_BACKTRACE_ENVVAR_NAME: RUST_FULL_BACKTRACE_KEYWORD}
     env.update(extra_env_vars)
-    return ServiceConfig(
-        image=image,
-        ports=used_ports,
-        public_ports=public_ports,
-        cmd=cmd,
-        files=files,
-        env_vars=env,
-        private_ip_address_placeholder=constants.PRIVATE_IP_ADDRESS_PLACEHOLDER,
-        ready_conditions=cl_node_ready_conditions.get_ready_conditions(
+    config_args = {
+        "image": image,
+        "ports": used_ports,
+        "public_ports": public_ports,
+        "cmd": cmd,
+        "files": files,
+        "env_vars": env,
+        "private_ip_address_placeholder": constants.PRIVATE_IP_ADDRESS_PLACEHOLDER,
+        "ready_conditions": cl_node_ready_conditions.get_ready_conditions(
             constants.HTTP_PORT_ID
         ),
-        min_cpu=cl_min_cpu,
-        max_cpu=cl_max_cpu,
-        min_memory=cl_min_mem,
-        max_memory=cl_max_mem,
-        labels=shared_utils.label_maker(
+        "labels": shared_utils.label_maker(
             constants.CL_TYPE.lighthouse,
             constants.CLIENT_TYPES.cl,
             image,
             el_context.client_name,
             extra_labels,
         ),
-        tolerations=tolerations,
-        node_selectors=node_selectors,
-    )
+        "tolerations": tolerations,
+        "node_selectors": node_selectors,
+    }
+
+    if cl_min_cpu > 0:
+        config_args["min_cpu"] = cl_min_cpu
+
+    if cl_max_cpu > 0:
+        config_args["max_cpu"] = cl_max_cpu
+
+    if cl_min_mem > 0:
+        config_args["min_memory"] = cl_min_mem
+
+    if cl_max_mem > 0:
+        config_args["max_memory"] = cl_max_mem
+
+    return ServiceConfig(**config_args)
 
 
 def new_lighthouse_launcher(el_cl_genesis_data, jwt_file, network_params):