[develop] [POC] Add new DebugLevel DevSetting parameter

cli/src/pcluster/config/cluster_config.py

@@ -1121,6 +1121,7 @@ def __init__(
         instance_types_data: str = None,
         timeouts: Timeouts = None,
         compute_startup_time_metric_enabled: bool = None,
+        debug_level: str = None,
         **kwargs,
     ):
         super().__init__(**kwargs)
@@ -1131,6 +1132,7 @@ def __init__(
         self.compute_startup_time_metric_enabled = Resource.init_param(
             compute_startup_time_metric_enabled, default=False
         )
+        self.debug_level = Resource.init_param(debug_level, default="info")
 
     def _register_validators(self, context: ValidatorContext = None):
         super()._register_validators(context)

cli/src/pcluster/resources/compute_node/user_data.sh

@@ -104,7 +104,8 @@ write_files:
           "head_node_private_ip": "${HeadNodePrivateIp}",
           "directory_service": {
             "enabled": "${DirectoryServiceEnabled}"
-          }
+          },
+          "debug_level": "${DebugLevel}"
         }
       }
   - path: /etc/chef/client.rb
@@ -125,11 +126,14 @@ write_files:
       function error_exit
       {
         echo "Bootstrap failed with error: $1"
-        # wait logs flush before signaling the failure
-        sleep 10
-        # TODO: add possibility to override this behavior and keep the instance for debugging
-        shutdown -h now
-        exit 1
+        if [ "${DebugLevel}" != "info" ]; then
+          echo "Skipping termination because debug_level is set to ${DebugLevel}"
+        else
+          # wait logs flush before signaling the failure
+          sleep 10
+          shutdown -h now
+          exit 1
+        fi
       }
       function vendor_cookbook
       {
@@ -214,11 +218,11 @@ write_files:
       jq --argfile f1 /tmp/dna.json --argfile f2 /tmp/extra.json -n '$f1 * $f2' > /etc/chef/dna.json || ( echo "jq not installed or invalid extra_json"; cp /tmp/dna.json /etc/chef/dna.json)
       {
         pushd /etc/chef &&
-        cinc-client --local-mode --config /etc/chef/client.rb --log_level info --force-formatter --no-color --chef-zero-port 8889 --json-attributes /etc/chef/dna.json --override-runlist aws-parallelcluster-entrypoints::init &&
+        cinc-client --local-mode --config /etc/chef/client.rb --log_level ${DebugLevel} --force-formatter --no-color --chef-zero-port 8889 --json-attributes /etc/chef/dna.json --override-runlist aws-parallelcluster-entrypoints::init &&
         /opt/parallelcluster/scripts/fetch_and_run -preinstall &&
-        cinc-client --local-mode --config /etc/chef/client.rb --log_level info --force-formatter --no-color --chef-zero-port 8889 --json-attributes /etc/chef/dna.json --override-runlist aws-parallelcluster-entrypoints::config &&
+        cinc-client --local-mode --config /etc/chef/client.rb --log_level ${DebugLevel} --force-formatter --no-color --chef-zero-port 8889 --json-attributes /etc/chef/dna.json --override-runlist aws-parallelcluster-entrypoints::config &&
         /opt/parallelcluster/scripts/fetch_and_run -postinstall &&
-        cinc-client --local-mode --config /etc/chef/client.rb --log_level info --force-formatter --no-color --chef-zero-port 8889 --json-attributes /etc/chef/dna.json --override-runlist aws-parallelcluster-entrypoints::finalize &&
+        cinc-client --local-mode --config /etc/chef/client.rb --log_level ${DebugLevel} --force-formatter --no-color --chef-zero-port 8889 --json-attributes /etc/chef/dna.json --override-runlist aws-parallelcluster-entrypoints::finalize &&
         popd
       } || error_exit 'Failed to run bootstrap recipes. If --norollback was specified, check /var/log/cfn-init.log and /var/log/cloud-init-output.log.'
 

cli/src/pcluster/schemas/cluster_schema.py

@@ -1070,6 +1070,7 @@ class ClusterDevSettingsSchema(BaseDevSettingsSchema):
     instance_types_data = fields.Str(metadata={"update_policy": UpdatePolicy.SUPPORTED})
     timeouts = fields.Nested(TimeoutsSchema, metadata={"update_policy": UpdatePolicy.SUPPORTED})
     compute_startup_time_metric_enabled = fields.Bool(metadata={"update_policy": UpdatePolicy.SUPPORTED})
+    debug_level = fields.Str(metadata={"update_policy": UpdatePolicy.SUPPORTED})
 
     @post_load
     def make_resource(self, data, **kwargs):

cli/src/pcluster/templates/cdk_builder_utils.py

@@ -87,6 +87,7 @@ def get_common_user_data_env(node: Union[HeadNode, SlurmQueue, LoginNodesPool],
         "CookbookVersion": COOKBOOK_PACKAGES_VERSIONS["cookbook"],
         "ChefVersion": COOKBOOK_PACKAGES_VERSIONS["chef"],
         "BerkshelfVersion": COOKBOOK_PACKAGES_VERSIONS["berkshelf"],
+        "DebugLevel": config.dev_settings.debug_level,
     }
 
 

cli/src/pcluster/templates/cluster_stack.py

@@ -1362,7 +1362,8 @@ def _add_head_node(self):
                 "commands": {
                     "chef": {
                         "command": (
-                            "cinc-client --local-mode --config /etc/chef/client.rb --log_level info "
+                            "cinc-client --local-mode --config /etc/chef/client.rb "
+                            f"--log_level {self.config.dev_settings.debug_level} "
                             "--logfile /var/log/chef-client.log --force-formatter --no-color "
                             "--chef-zero-port 8889 --json-attributes /etc/chef/dna.json "
                             "--override-runlist aws-parallelcluster-entrypoints::init"
@@ -1378,7 +1379,8 @@ def _add_head_node(self):
                 "commands": {
                     "chef": {
                         "command": (
-                            "cinc-client --local-mode --config /etc/chef/client.rb --log_level info "
+                            "cinc-client --local-mode --config /etc/chef/client.rb "
+                            f"--log_level {self.config.dev_settings.debug_level} "
                             "--logfile /var/log/chef-client.log --force-formatter --no-color "
                             "--chef-zero-port 8889 --json-attributes /etc/chef/dna.json "
                             "--override-runlist aws-parallelcluster-entrypoints::config"
@@ -1394,7 +1396,8 @@ def _add_head_node(self):
                 "commands": {
                     "chef": {
                         "command": (
-                            "cinc-client --local-mode --config /etc/chef/client.rb --log_level info "
+                            "cinc-client --local-mode --config /etc/chef/client.rb "
+                            f"--log_level {self.config.dev_settings.debug_level} "
                             "--logfile /var/log/chef-client.log --force-formatter --no-color "
                             "--chef-zero-port 8889 --json-attributes /etc/chef/dna.json "
                             "--override-runlist aws-parallelcluster-entrypoints::finalize"
@@ -1414,7 +1417,8 @@ def _add_head_node(self):
                     "chef": {
                         "command": (
                             ". /etc/profile.d/pcluster.sh; "
-                            "cinc-client --local-mode --config /etc/chef/client.rb --log_level info"
+                            "cinc-client --local-mode --config /etc/chef/client.rb "
+                            f"--log_level {self.config.dev_settings.debug_level}"
                             " --logfile /var/log/chef-client.log --force-formatter --no-color"
                             " --chef-zero-port 8889 --json-attributes /etc/chef/dna.json"
                             " --override-runlist aws-parallelcluster-entrypoints::update &&"

cli/src/pcluster/templates/queues_stack.py

@@ -198,7 +198,9 @@ def _add_compute_resource_launch_template(
                 instance_market_options=self._launch_template_builder.get_instance_market_options(
                     queue, compute_resource
                 ),
-                instance_initiated_shutdown_behavior="terminate",
+                instance_initiated_shutdown_behavior="stop"
+                if self._config.dev_settings.debug_level != "info"
+                else "terminate",
                 capacity_reservation_specification=self._launch_template_builder.get_capacity_reservation(
                     queue,
                     compute_resource,