From 79f18160e7a7d4172acdece2c7b22ad710a3b275 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Henrik=20tom=20W=C3=B6rden?= <h.tomwoerden@indiscale.com>
Date: Thu, 9 Jan 2025 11:23:34 +0100
Subject: [PATCH 1/4] TST: add performance test
---
performance-tests/test.py | 121 ++++++++++++++++++++++++++++++++++++++
pyproject.toml | 2 +
2 files changed, 123 insertions(+)
create mode 100644 performance-tests/test.py
diff --git a/performance-tests/test.py b/performance-tests/test.py
new file mode 100644
index 0000000..6409fb9
--- /dev/null
+++ b/performance-tests/test.py
@@ -0,0 +1,121 @@
+"""
+Tests CPU and Memory usage of RuQaD
+"""
+import cProfile, pstats, io
+from pstats import SortKey
+from time import sleep,time
+from tempfile import TemporaryDirectory
+from datetime import datetime, timezone
+from pathlib import Path
+
+from memory_profiler import profile as mprofile
+
+from ruqad.qualitycheck import QualityChecker
+from ruqad.kadi import collect_records_created_after, download_eln_for, KadiManager
+from ruqad.crawler import trigger_crawler
+import os
+import shutil
+from memory_profiler import memory_usage
+
+SKIP_QUALITY_CHECK = os.getenv("SKIP_QUALITY_CHECK") is not None
+KADIARGS = {
+ "host": os.environ['KADIHOST'],
+ "pat": os.environ['KADITOKEN'],
+}
+
+"""
+call this file to check memory and cpu usage of the RuQaD demonstrator
+"""
+
+def _run(n=-1):
+ cut_off_date = datetime.fromisoformat("1990-01-01 02:34:42.484312+00:00")
+ with KadiManager(**KADIARGS) as manager:
+ print(f"Checking for records created after {cut_off_date}...")
+ rec_ids = collect_records_created_after(manager, cut_off_date)
+
+ if len(rec_ids) == 0:
+ print("no new recs")
+ if n!=-1:
+ rec_ids = rec_ids[:n]
+ for rid in rec_ids:
+ with TemporaryDirectory(delete=False) as cdir:
+ eln_file = os.path.join(cdir, "export.eln")
+ download_eln_for(manager, rid, path=eln_file)
+ print(f"Downlaoded {eln_file}")
+ if SKIP_QUALITY_CHECK:
+ print("Found env 'SKIP_QUALITY_CHECK', skipping quality check")
+ else:
+ qc = QualityChecker()
+ qc.check(filename=eln_file, target_dir=cdir)
+ print(f"Quality check done. {os.listdir(cdir)}")
+ # trigger crawler on dir
+ remote_dir_path = os.path.join(cdir, "ruqad", str(rid))
+ os.makedirs(remote_dir_path)
+ if os.path.exists(os.path.join(cdir, "artifacts.zip")):
+ shutil.move(os.path.join(cdir, "artifacts.zip"),
+ os.path.join(remote_dir_path, "report.zip"))
+ #else:
+ # Path(os.path.join(remote_dir_path, "report.zip")).touch()
+ shutil.move(os.path.join(cdir, "export.eln"),
+ os.path.join(remote_dir_path, "export.eln"))
+ trigger_crawler(target_dir=cdir)
+
+def test_memory():
+ # test that maximum memory usage is below 1GB
+ assert 1000>max(memory_usage((_run, [10], {})))
+
+def test_cpu():
+ pr = cProfile.Profile()
+ pr.enable()
+ _run(n=1)
+ pr.disable()
+ s = io.StringIO()
+ sortby = SortKey.CUMULATIVE
+ ps = pstats.Stats(pr, stream=s).sort_stats(sortby)
+ ps.print_stats(10)
+ print(s.getvalue())
+ ps.print_stats("ruqad", 10)
+ ps.print_stats("crawler", 10)
+ print(s.getvalue())
+
+def test_runtime_eln_download():
+ cut_off_date = datetime.fromisoformat("1990-01-01 02:34:42.484312+00:00")
+ with KadiManager(**KADIARGS) as manager:
+ rec_ids = collect_records_created_after(manager, cut_off_date)
+
+ if len(rec_ids) == 0:
+ print("no new recs")
+ with TemporaryDirectory(delete=False) as cdir:
+ eln_file = os.path.join(cdir, "export.eln")
+ start = time()
+ download_eln_for(manager, rec_ids[0], path=eln_file)
+ stop = time()
+
+ print(f"time for downloading eln: {stop-start:.2f} s")
+
+def test_runtime_crawler():
+ cut_off_date = datetime.fromisoformat("1990-01-01 02:34:42.484312+00:00")
+ with KadiManager(**KADIARGS) as manager:
+ rec_ids = collect_records_created_after(manager, cut_off_date)
+
+ if len(rec_ids) == 0:
+ print("no new recs")
+ with TemporaryDirectory(delete=False) as cdir:
+ eln_file = os.path.join(cdir, "export.eln")
+ download_eln_for(manager, rec_ids[0], path=eln_file)
+ # trigger crawler on dir
+ remote_dir_path = os.path.join(cdir, "ruqad", str(rec_ids[0]))
+ os.makedirs(remote_dir_path)
+ shutil.move(os.path.join(cdir, "export.eln"),
+ os.path.join(remote_dir_path, "export.eln"))
+ start = time()
+ trigger_crawler(target_dir=cdir)
+ stop = time()
+
+ print(f"time for crawling eln: {stop-start:.2f} s")
+
+if __name__ == "__main__":
+ test_memory()
+ test_runtime_eln_download()
+ test_cpu()
+ test_runtime_crawler()
diff --git a/pyproject.toml b/pyproject.toml
index fd35267..c81a9cf 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -47,6 +47,8 @@ test = [
"tox",
"pytest",
"pytest-cov",
+ "guppy3",
+ "memory_profiler",
]
all = [
"ruqad[dev]",
--
GitLab
From 42a4f27e5ed28a3166c0b8418daabeaa012efc4e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Henrik=20tom=20W=C3=B6rden?= <h.tomwoerden@indiscale.com>
Date: Thu, 9 Jan 2025 11:34:54 +0100
Subject: [PATCH 2/4] FIX: add ini for docker
---
docker/Dockerfile | 2 +-
docker/pylinkahead.ini | 36 ++++++++++++++++++++++++++++++++++++
2 files changed, 37 insertions(+), 1 deletion(-)
create mode 100644 docker/pylinkahead.ini
diff --git a/docker/Dockerfile b/docker/Dockerfile
index 638ffac..37b0aec 100644
--- a/docker/Dockerfile
+++ b/docker/Dockerfile
@@ -8,7 +8,7 @@ COPY ./pyproject.toml /ruqad/
# static configuration
COPY ./qualitycheck_config.toml /ruqad/
-COPY ./pylinkahead.ini /ruqad/
+COPY ./docker/pylinkahead.ini /ruqad/
# Installing the package
WORKDIR /ruqad/
diff --git a/docker/pylinkahead.ini b/docker/pylinkahead.ini
new file mode 100644
index 0000000..a7a4fc0
--- /dev/null
+++ b/docker/pylinkahead.ini
@@ -0,0 +1,36 @@
+# The INI file must be located either in
+# - $CWD/pylinkahead.ini
+# - $HOME/.pylinkahead.ini
+# - the location given in the env variable PYLINKAHEADINI
+
+[Connection]
+# url=https://demo.indiscale.com/
+url = http://localhost:80/provider/linkahead/
+
+## If this option is set, the SSL certificate will be ignored. Use with care!
+#ssl_insecure=1
+
+username=admin
+
+## The password input method can be chosen with the `password_method` setting,
+## which by default is set to `plain`.
+##
+## DEFAULT: the password method is `plain`, now the password must be saved as
+## plain text.
+password_method=input
+# password=caosdb
+
+
+## OR: `input`: username is optional, password is entered by the user directly
+# password_method=input
+
+## OR: `pass`: password is retrieved from the "pass" password manager
+# password_method=pass
+# password_identifier=...
+
+## OR: `keyring`: using the system keyring/wallet (macOS, GNOME, KDE, Windows)
+## requires installation of the keyring python package:
+## pip install keyring
+# password_method=keyring
+
+timeout=10000
--
GitLab
From 655d9468d00c0ce37a1bac0ffd91e45bdcd488d5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Henrik=20tom=20W=C3=B6rden?= <h.tomwoerden@indiscale.com>
Date: Thu, 9 Jan 2025 11:39:55 +0100
Subject: [PATCH 3/4] MAINT: remove unused file
---
docker/pylinkahead.ini | 36 ------------------------------------
1 file changed, 36 deletions(-)
delete mode 100644 docker/pylinkahead.ini
diff --git a/docker/pylinkahead.ini b/docker/pylinkahead.ini
deleted file mode 100644
index a7a4fc0..0000000
--- a/docker/pylinkahead.ini
+++ /dev/null
@@ -1,36 +0,0 @@
-# The INI file must be located either in
-# - $CWD/pylinkahead.ini
-# - $HOME/.pylinkahead.ini
-# - the location given in the env variable PYLINKAHEADINI
-
-[Connection]
-# url=https://demo.indiscale.com/
-url = http://localhost:80/provider/linkahead/
-
-## If this option is set, the SSL certificate will be ignored. Use with care!
-#ssl_insecure=1
-
-username=admin
-
-## The password input method can be chosen with the `password_method` setting,
-## which by default is set to `plain`.
-##
-## DEFAULT: the password method is `plain`, now the password must be saved as
-## plain text.
-password_method=input
-# password=caosdb
-
-
-## OR: `input`: username is optional, password is entered by the user directly
-# password_method=input
-
-## OR: `pass`: password is retrieved from the "pass" password manager
-# password_method=pass
-# password_identifier=...
-
-## OR: `keyring`: using the system keyring/wallet (macOS, GNOME, KDE, Windows)
-## requires installation of the keyring python package:
-## pip install keyring
-# password_method=keyring
-
-timeout=10000
--
GitLab
From e92628537ce3d970394536a7a0ad2b7317a66c12 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Henrik=20tom=20W=C3=B6rden?= <h.tomwoerden@indiscale.com>
Date: Fri, 10 Jan 2025 13:19:09 +0100
Subject: [PATCH 4/4] TST: write output to file
---
performance-tests/test.py | 41 ++++++++++++++++++++-------------------
1 file changed, 21 insertions(+), 20 deletions(-)
diff --git a/performance-tests/test.py b/performance-tests/test.py
index 6409fb9..82d5b70 100644
--- a/performance-tests/test.py
+++ b/performance-tests/test.py
@@ -1,21 +1,22 @@
"""
Tests CPU and Memory usage of RuQaD
"""
-import cProfile, pstats, io
-from pstats import SortKey
-from time import sleep,time
-from tempfile import TemporaryDirectory
+import cProfile
+import io
+import os
+import pstats
+import shutil
from datetime import datetime, timezone
from pathlib import Path
+from pstats import SortKey
+from tempfile import TemporaryDirectory
+from time import sleep, time
+from memory_profiler import memory_usage
from memory_profiler import profile as mprofile
-
-from ruqad.qualitycheck import QualityChecker
-from ruqad.kadi import collect_records_created_after, download_eln_for, KadiManager
from ruqad.crawler import trigger_crawler
-import os
-import shutil
-from memory_profiler import memory_usage
+from ruqad.kadi import KadiManager, collect_records_created_after, download_eln_for
+from ruqad.qualitycheck import QualityChecker
SKIP_QUALITY_CHECK = os.getenv("SKIP_QUALITY_CHECK") is not None
KADIARGS = {
@@ -69,14 +70,12 @@ def test_cpu():
pr.enable()
_run(n=1)
pr.disable()
- s = io.StringIO()
- sortby = SortKey.CUMULATIVE
- ps = pstats.Stats(pr, stream=s).sort_stats(sortby)
- ps.print_stats(10)
- print(s.getvalue())
- ps.print_stats("ruqad", 10)
- ps.print_stats("crawler", 10)
- print(s.getvalue())
+ with open("performance.txt", 'a') as s:
+ sortby = SortKey.CUMULATIVE
+ ps = pstats.Stats(pr, stream=s).sort_stats(sortby)
+ ps.print_stats(10)
+ ps.print_stats("ruqad", 10)
+ ps.print_stats("crawler", 10)
def test_runtime_eln_download():
cut_off_date = datetime.fromisoformat("1990-01-01 02:34:42.484312+00:00")
@@ -91,7 +90,8 @@ def test_runtime_eln_download():
download_eln_for(manager, rec_ids[0], path=eln_file)
stop = time()
- print(f"time for downloading eln: {stop-start:.2f} s")
+ with open("performance.txt", 'a') as s:
+ s.write(f"time for downloading eln: {stop-start:.2f} s")
def test_runtime_crawler():
cut_off_date = datetime.fromisoformat("1990-01-01 02:34:42.484312+00:00")
@@ -112,7 +112,8 @@ def test_runtime_crawler():
trigger_crawler(target_dir=cdir)
stop = time()
- print(f"time for crawling eln: {stop-start:.2f} s")
+ with open("performance.txt", 'a') as s:
+ s.write(f"time for crawling eln: {stop-start:.2f} s")
if __name__ == "__main__":
test_memory()
--
GitLab