Added cmdline argument to Java app ID based on a regex filter (#328)

* Added cmdline argument to Java app ID based on a regex filter

* import issue fix

* import issue fix

Author: liorgorb

gprofiler/main.py

@@ -17,6 +17,8 @@
 from typing import Any, Iterable, Optional, Type, cast
 
 import configargparse
+
+from gprofiler.metadata.enrichment import EnrichmentOptions
 from granulate_utils.linux.ns import is_running_in_init_pid
 from granulate_utils.linux.process import is_process_running
 from psutil import NoSuchProcess, Process
@@ -28,7 +30,8 @@
 from gprofiler.exceptions import APIError, SystemProfilerInitFailure
 from gprofiler.gprofiler_types import ProcessToStackSampleCounters, UserArgs, positive_integer
 from gprofiler.log import RemoteLogsHandler, initial_root_logger_setup
-from gprofiler.merge import EnrichmentOptions, concatenate_profiles, merge_profiles
+from gprofiler.merge import concatenate_profiles, merge_profiles
+from gprofiler.metadata.application_identifiers import set_enrichment_options
 from gprofiler.metadata.metadata_collector import get_current_metadata, get_static_metadata
 from gprofiler.metadata.metadata_type import Metadata
 from gprofiler.metadata.system_metadata import get_hostname, get_run_mode, get_static_system_info
@@ -543,6 +546,14 @@ def parse_cmd_args() -> Any:
         help="Disable identification of applications by heuristics",
     )
 
+    parser.add_argument(
+        "--app-id-args-filter",
+        action="append",
+        default=list(),
+        dest="app_id_args_filters",
+        help="A regex based filter for adding relevant arguments to the app id",
+    )
+
     parser.add_argument(
         "--disable-application-metadata",
         action="store_false",
@@ -734,9 +745,12 @@ def main() -> None:
             profile_api_version=args.profile_api_version,
             container_names=args.container_names,
             application_identifiers=args.identify_applications,
+            application_identifier_args_filters=args.app_id_args_filters,
             application_metadata=args.application_metadata,
         )
 
+        set_enrichment_options(enrichment_options)
+
         gprofiler = GProfiler(
             args.output_dir,
             args.flamegraph,

gprofiler/merge.py

@@ -16,6 +16,7 @@
 from gprofiler.log import get_logger_adapter
 from gprofiler.metadata import application_identifiers
 from gprofiler.metadata.application_metadata import get_application_metadata
+from gprofiler.metadata.enrichment import EnrichmentOptions
 from gprofiler.metadata.metadata_type import Metadata
 from gprofiler.system_metrics import Metrics
 
@@ -33,18 +34,6 @@
 FRAME_REGEX = re.compile(r"^\s*[0-9a-f]+ (.*?) \((.*)\)$")
 
 
-@dataclass
-class EnrichmentOptions:
-    """
-    Profile enrichment options.
-    """
-
-    profile_api_version: str  # profile protocol version. v1 does not support container_names and application_metadata.
-    container_names: bool  # Include container names for each stack in result profile
-    application_identifiers: bool  # Attempt to produce & include appid frames for each stack in result profile
-    application_metadata: bool  # Include specialized metadata per application, e.g for Python - the Python version
-
-
 def parse_one_collapsed(collapsed: str, add_comm: Optional[str] = None) -> StackToSampleCount:
     """
     Parse a stack-collapsed listing.

gprofiler/metadata/application_identifiers.py

@@ -8,6 +8,7 @@
 from abc import ABCMeta, abstractmethod
 from typing import List, Optional, TextIO, Tuple, Union
 
+from gprofiler.metadata.enrichment import EnrichmentOptions
 from granulate_utils.linux.ns import resolve_host_path, resolve_proc_root_links
 from psutil import NoSuchProcess, Process
 
@@ -91,6 +92,8 @@ def _append_file_to_proc_wd(process: Process, file_path: str) -> str:
 
 
 class _ApplicationIdentifier(metaclass=ABCMeta):
+    enrichment_options: Optional[EnrichmentOptions] = None
+
     @abstractmethod
     def get_application_name(self, process: Process) -> Optional[str]:
         pass
@@ -262,11 +265,25 @@ def get_application_name(self, process: Process) -> Optional[str]:
             return None
 
         try:
-            java_properties = run_process([jattach_path(), str(process.pid), "properties"]).stdout.decode()
+            java_properties = run_process([jattach_path(), str(process.pid), "jcmd", "VM.command_line"]).stdout.decode()
+            java_command = None
+            java_args = []
             for line in java_properties.splitlines():
-                if line.startswith("sun.java.command"):
-                    app_id = line[line.find("=") + 1 :].split(" ", 1)[0]
-                    return f"java: {app_id}"
+                if line.startswith("jvm_args:"):
+                    if (
+                        self.enrichment_options is not None
+                        and self.enrichment_options.application_identifier_args_filters
+                    ):
+                        for arg in line[line.find(":") + 1 :].strip().split(" "):
+                            if any(
+                                re.search(flag_filter, arg)
+                                for flag_filter in self.enrichment_options.application_identifier_args_filters
+                            ):
+                                java_args.append(arg)
+                if line.startswith("java_command:"):
+                    java_command = line[line.find(":") + 1 :].strip().split(" ", 1)[0]
+            if java_command:
+                return f"java: {java_command}{' (' + ' '.join(java_args) + ')' if java_args else ''}"
         except CalledProcessError as e:
             _logger.warning(f"Couldn't get Java properties for process {process.pid}: {e.stderr}")
 
@@ -286,6 +303,10 @@ def get_application_name(self, process: Process) -> Optional[str]:
 ]
 
 
+def set_enrichment_options(enrichment_options: EnrichmentOptions) -> None:
+    _ApplicationIdentifier.enrichment_options = enrichment_options
+
+
 def get_application_name(process: Union[int, Process]) -> Optional[str]:
     """
     Tries to identify the application running in a given process, application identification is fully heuristic,

gprofiler/metadata/enrichment.py

@@ -0,0 +1,16 @@
+from typing import List
+
+from dataclasses import dataclass
+
+
+@dataclass
+class EnrichmentOptions:
+    """
+    Profile enrichment options.
+    """
+
+    profile_api_version: str  # profile protocol version. v1 does not support container_names and application_metadata.
+    container_names: bool  # Include container names for each stack in result profile
+    application_identifiers: bool  # Attempt to produce & include appid frames for each stack in result profile
+    application_identifier_args_filters: List[str]  # A list of regex filters to add cmdline arguments to the app id
+    application_metadata: bool  # Include specialized metadata per application, e.g for Python - the Python version