Merge pull request #555 from Kaufi-Jonas/j-optimize_perf

speed up slow build times for ./configure based projects

.isort.cfg

@@ -1,5 +1,5 @@
 [settings]
-known_third_party = attr,dill,faker,git,jinja2,mock,parse,pathos,pkg_resources,plumbum,psutil,pygit2,pygtrie,pyparsing,pytest,pytest_git,result,rich,schema,setuptools,six,sqlalchemy,yaml
+known_third_party = attr,dill,faker,git,jinja2,mock,parse,plumbum,psutil,pygtrie,pyparsing,pytest,pytest_git,result,rich,schema,setuptools,six,sqlalchemy,yaml
 multi_line_output=3
 use_parentheses = True
 include_trailing_comma: True

.pylintrc

@@ -263,4 +263,4 @@ int-import-graph=
 
 [EXCEPTIONS]
 
-overgeneral-exceptions=Exception
+overgeneral-exceptions=builtins.Exception

benchbuild/__init__.py

@@ -28,8 +28,8 @@
 
 def __init__() -> None:
     """Initialize all plugins and settings."""
-    __PLUGINS__.discover()
-    __SETTINGS__.setup_config(CFG)
+    if __PLUGINS__.discover():
+        __SETTINGS__.setup_config(CFG)
 
 
 __init__()

benchbuild/environments/adapters/podman.py

@@ -5,7 +5,6 @@
 from plumbum import local, ProcessExecutionError
 from result import Result, Err, Ok
 from rich import print
-from rich.markdown import Markdown
 
 from benchbuild.environments.adapters import buildah
 from benchbuild.environments.adapters.common import (
@@ -173,6 +172,9 @@ def _create(
                     '--mount', f'type=bind,src={source},target={target}']
 
         if interactive:
+            # pylint: disable=import-outside-toplevel
+            from rich.markdown import Markdown
+
             entrypoint = buildah.find_entrypoint(image.name)
             print(
                 Markdown(

benchbuild/environments/service_layer/debug.py

@@ -1,6 +1,5 @@
 from plumbum import ProcessExecutionError
 from rich import print
-from rich.markdown import Markdown
 
 from benchbuild.environments.adapters.common import bb_buildah
 from benchbuild.environments.domain import events
@@ -25,6 +24,8 @@ def debug_image_kept(
     """
     Spawn a debug session of the kept image and provide diagnostics.
     """
+    # pylint: disable=import-outside-toplevel
+    from rich.markdown import Markdown
     with uow:
         container = uow.create(event.image_name, event.failed_image_name)
         if container is None:

benchbuild/extensions/compiler.py

@@ -6,11 +6,12 @@
 from plumbum.commands.base import BoundCommand
 
 from benchbuild.extensions import base
+from benchbuild.settings import CFG
 from benchbuild.utils import db, run
 
 if TYPE_CHECKING:
-    from benchbuild.project import Project
     from benchbuild.experiment import Experiment
+    from benchbuild.project import Project
 
 LOG = logging.getLogger(__name__)
 
@@ -65,9 +66,10 @@ def __call__(
                         default_flow_style=False
                     )
                 )
-                db.persist_config(
-                    run_info.db_run, run_info.session, self.config
-                )
+                if CFG["db"]["enabled"]:
+                    db.persist_config(
+                        run_info.db_run, run_info.session, self.config
+                    )
 
             if run_info.has_failed:
                 with run.track_execution(

benchbuild/extensions/run.py

@@ -5,6 +5,7 @@
 from plumbum import local
 
 from benchbuild.extensions import base
+from benchbuild.settings import CFG
 from benchbuild.utils import db, run
 from benchbuild.utils.settings import get_number_of_jobs
 
@@ -45,9 +46,10 @@ def __call__(self, binary_command, *args, **kwargs):
                 )
                 self.config['baseline'] = \
                     os.getenv("BB_IS_BASELINE", "False")
-                db.persist_config(
-                    run_info.db_run, run_info.session, self.config
-                )
+                if CFG["db"]["enabled"]:
+                    db.persist_config(
+                        run_info.db_run, run_info.session, self.config
+                    )
         res = self.call_next(binary_command, *args, **kwargs)
         res.append(run_info)
         return res
@@ -69,6 +71,7 @@ def __init__(self, *extensions, limit="10m", **kwargs):
         self.limit = limit
 
     def __call__(self, binary_command, *args, **kwargs):
+        # pylint: disable=import-outside-toplevel
         from benchbuild.utils.cmd import timeout
         return self.call_next(
             timeout[self.limit, binary_command], *args, **kwargs
@@ -83,8 +86,6 @@ class SetThreadLimit(base.Extension):
     """
 
     def __call__(self, binary_command, *args, **kwargs):
-        from benchbuild.settings import CFG
-
         config = self.config
         if config is not None and 'jobs' in config.keys():
             jobs = get_number_of_jobs(config)

benchbuild/extensions/time.py

@@ -4,6 +4,7 @@
 import parse
 
 from benchbuild.extensions import base
+from benchbuild.settings import CFG
 from benchbuild.utils import db
 from benchbuild.utils.cmd import time
 
@@ -20,6 +21,10 @@ def __call__(self, binary_command, *args, may_wrap=True, **kwargs):
 
         def handle_timing(run_infos):
             """Takes care of the formating for the timing statistics."""
+            if not CFG["db"]["enabled"]:
+                return run_infos
+
+            # pylint: disable=import-outside-toplevel
             from benchbuild.utils import schema as s
 
             session = s.Session()

benchbuild/plugins.py

@@ -23,15 +23,19 @@
 LOG = logging.getLogger(__name__)
 
 
-def discover() -> None:
+def discover() -> bool:
     """Import all plugins listed in our configuration."""
+    something_imported = False
     if CFG["plugins"]["autoload"]:
         experiment_plugins = CFG["plugins"]["experiments"].value
         project_plugins = CFG["plugins"]["projects"].value
 
         for plugin in itertools.chain(experiment_plugins, project_plugins):
+            something_imported = True
             try:
                 importlib.import_module(plugin)
             except ImportError as import_error:
                 LOG.error("Could not find '%s'", import_error.name)
                 LOG.debug("ImportError: %s", import_error)
+
+    return something_imported

benchbuild/project.py

@@ -336,7 +336,8 @@ def __default_primary_source(self) -> str:  # pylint: disable=unused-private-mem
     runtime_extension = attr.ib(default=None)
 
     def __attrs_post_init__(self) -> None:
-        db.persist_project(self)
+        if CFG["db"]["enabled"]:
+            db.persist_project(self)
 
         # select container image
         if isinstance(type(self).CONTAINER, ContainerImage):
@@ -345,7 +346,7 @@ def __attrs_post_init__(self) -> None:
             )
         else:
             primary_source = primary(*self.SOURCE)
-            if isinstance(primary_source,source.BaseVersionFilter):
+            if isinstance(primary_source, source.BaseVersionFilter):
                 primary_source = primary_source.child
             if not isinstance(primary_source, Git):
                 raise AssertionError(
@@ -361,7 +362,6 @@ def __attrs_post_init__(self) -> None:
                         self.container = copy.deepcopy(image)
                         break
 
-
     def clean(self) -> None:
         """Clean the project build directory."""
         builddir_p = local.path(self.builddir)

benchbuild/res/wrapping/run_compiler.py.inc

@@ -6,6 +6,12 @@ os.environ["OPENBLAS_NUM_THREADS"] = "4"
 
 from plumbum import TEE, local
 
+# Performance optimization for benchbuild: don't import any experiments or
+# projects. Everything necessary should be imported when loading (unpickling)
+# the project and the compiler.
+os.environ["BB_PLUGINS_AUTOLOAD"] = "False"
+
+from benchbuild.settings import CFG
 from benchbuild.utils import log
 from benchbuild.utils.db import persist_project
 from benchbuild.utils.run import exit_code_from_run_infos
@@ -24,7 +30,8 @@ def update_project(argv):
                 name = project_p.basename
                 break
         PROJECT.name = name
-        persist_project(PROJECT)
+        if CFG["db"]["enabled"]:
+            persist_project(PROJECT)
 
 
 def main(argv):

benchbuild/res/wrapping/run_dynamic.py.inc

@@ -6,6 +6,11 @@ import sys
 
 from plumbum import TEE, local
 
+# Performance optimization for benchbuild: don't import any experiments or
+# projects. Everything necessary should be imported when loading (unpickling)
+# the project.
+os.environ["BB_PLUGINS_AUTOLOAD"] = "False"
+
 from benchbuild.utils import log
 from benchbuild.utils.db import persist_project
 from benchbuild.utils.run import exit_code_from_run_infos

benchbuild/res/wrapping/run_static.py.inc

@@ -1,9 +1,15 @@
 #!{{ python|default("/usr/bin/env python3") }}
 #
+import os
 import sys
 
 from plumbum import TEE, local
 
+# Performance optimization for benchbuild: don't import any experiments or
+# projects. Everything necessary should be imported when loading (unpickling)
+# the project.
+os.environ["BB_PLUGINS_AUTOLOAD"] = "False"
+
 from benchbuild.utils import log
 from benchbuild.utils.run import exit_code_from_run_infos
 from benchbuild.utils.wrapping import load

benchbuild/settings.py

@@ -145,9 +145,13 @@
 }
 
 CFG['db'] = {
+    "enabled": {
+        "desc": "Whether the database is enabled.",
+        "default": False
+    },
     "connect_string": {
         "desc": "sqlalchemy connect string",
-        "default": "sqlite://"
+        "default": ""
     },
     "rollback": {
         "desc": "Rollback all operations after benchbuild completes.",
@@ -381,7 +385,7 @@
     "storage_driver": {
         "default": "vfs",
         "desc": "Storage driver for containers."
-     },
+    },
     "input": {
         "default": "container.tar.bz2",
         "desc": "Input container file/folder."

benchbuild/utils/actions.py

@@ -25,8 +25,6 @@
 import typing as tp
 from datetime import datetime
 
-import pathos.multiprocessing as mp
-import sqlalchemy as sa
 from plumbum import ProcessExecutionError
 
 from benchbuild import command, signals, source
@@ -250,7 +248,7 @@ def clean_mountpoints(root: str) -> None:
             root: All UnionFS-mountpoints under this directory will be
                   unmounted.
         """
-        import psutil
+        import psutil  # pylint: disable=import-outside-toplevel
 
         umount_paths = []
         real_root = os.path.realpath(root)
@@ -341,22 +339,27 @@ def __init__(
         self.experiment = experiment
 
     def __call__(self) -> StepResult:
-        group, session = run.begin_run_group(self.project, self.experiment)
-        signals.handlers.register(run.fail_run_group, group, session)
+        if CFG["db"]["enabled"]:
+            group, session = run.begin_run_group(self.project, self.experiment)
+            signals.handlers.register(run.fail_run_group, group, session)
         try:
             self.project.run_tests()
-            run.end_run_group(group, session)
+            if CFG["db"]["enabled"]:
+                run.end_run_group(group, session)
             self.status = StepResult.OK
         except ProcessExecutionError:
-            run.fail_run_group(group, session)
+            if CFG["db"]["enabled"]:
+                run.fail_run_group(group, session)
             self.status = StepResult.ERROR
             raise
         except KeyboardInterrupt:
-            run.fail_run_group(group, session)
+            if CFG["db"]["enabled"]:
+                run.fail_run_group(group, session)
             self.status = StepResult.ERROR
             raise
         finally:
-            signals.handlers.deregister(run.fail_run_group)
+            if CFG["db"]["enabled"]:
+                signals.handlers.deregister(run.fail_run_group)
 
         return self.status
 
@@ -444,6 +447,7 @@ def __init__(
     def begin_transaction(
         self,
     ) -> tp.Tuple["benchbuild.utils.schema.Experiment", tp.Any]:
+        import sqlalchemy as sa  # pylint: disable=import-outside-toplevel
         experiment, session = db.persist_experiment(self.experiment)
         if experiment.begin is None:
             experiment.begin = datetime.now()
@@ -467,6 +471,7 @@ def begin_transaction(
     def end_transaction(
         experiment: "benchbuild.utils.schema.Experiment", session: tp.Any
     ) -> None:
+        import sqlalchemy as sa  # pylint: disable=import-outside-toplevel
         try:
             experiment.end = max(experiment.end, datetime.now())
             session.add(experiment)
@@ -475,6 +480,9 @@ def end_transaction(
             LOG.error(inv_req)
 
     def __run_children(self, num_processes: int) -> tp.List[StepResult]:
+        # pylint: disable=import-outside-toplevel
+        import pathos.multiprocessing as mp
+
         results = []
         actions = self.actions
 
@@ -496,12 +504,14 @@ def __run_children(self, num_processes: int) -> tp.List[StepResult]:
     def __call__(self) -> StepResult:
         results = []
         session = None
-        experiment, session = self.begin_transaction()
+        if CFG["db"]["enabled"]:
+            experiment, session = self.begin_transaction()
         try:
             results = self.__run_children(int(CFG["parallel_processes"]))
         finally:
-            self.end_transaction(experiment, session)
-            signals.handlers.deregister(self.end_transaction)
+            if CFG["db"]["enabled"]:
+                self.end_transaction(experiment, session)
+                signals.handlers.deregister(self.end_transaction)
         self.status = max(results) if results else StepResult.OK
         return self.status
 
@@ -636,22 +646,27 @@ def __init__(
             ])
 
     def __call__(self) -> StepResult:
-        group, session = run.begin_run_group(self.project, self.experiment)
-        signals.handlers.register(run.fail_run_group, group, session)
+        if CFG["db"]["enabled"]:
+            group, session = run.begin_run_group(self.project, self.experiment)
+            signals.handlers.register(run.fail_run_group, group, session)
         try:
             self.status = max([workload() for workload in self.actions],
                               default=StepResult.OK)
-            run.end_run_group(group, session)
+            if CFG["db"]["enabled"]:
+                run.end_run_group(group, session)
         except ProcessExecutionError:
-            run.fail_run_group(group, session)
+            if CFG["db"]["enabled"]:
+                run.fail_run_group(group, session)
             self.status = StepResult.ERROR
             raise
         except KeyboardInterrupt:
-            run.fail_run_group(group, session)
+            if CFG["db"]["enabled"]:
+                run.fail_run_group(group, session)
             self.status = StepResult.ERROR
             raise
         finally:
-            signals.handlers.deregister(run.fail_run_group)
+            if CFG["db"]["enabled"]:
+                signals.handlers.deregister(run.fail_run_group)
 
         return self.status