Release 0.3.6 (2024-10-08)

build_analytics:
  buildsystem#360 Added src and x86_64_v2 arches

build_analytics/build_analytics/api_client.py

@@ -27,6 +27,8 @@ class APIclient():
         self.api_root = api_root
         self.jwt = jwt
         self.timeout = timeout
+        # will be set at first call of __send_request
+        self.session: Optional[requests.Session] = None
 
     def get_builds(self, page_num: int = 1) -> List[Build]:
         ep = '/api/v1/builds'
@@ -34,8 +36,7 @@ class APIclient():
         params = {'pageNumber': page_num}
         headers = {'accept': 'appilication/json'}
 
-        response = requests.get(
-            url, params=params, headers=headers, timeout=self.timeout)
+        response = self.__send_request(url, 'get', params, headers)
         response.raise_for_status()
 
         result = []
@@ -54,7 +55,7 @@ class APIclient():
         ep = f'/api/v1/builds/{build_id}'
         url = urljoin(self.api_root, ep)
         headers = {'accept': 'application/json'}
-        response = requests.get(url, headers=headers, timeout=self.timeout)
+        response = self.__send_request(url, 'get', headers=headers)
 
         if response.status_code == 404:
             return None
@@ -228,3 +229,31 @@ class APIclient():
                 start_ts = stat.start_ts
 
         return start_ts
+
+    def __send_request(self,
+                       url: str,
+                       method: str,
+                       params: Optional[Dict[str, Any]] = None,
+                       headers: Optional[Dict[str, Any]] = None,
+                       ) -> requests.Response:
+        """
+        Simple wrapper around requests.get/posts.. methods 
+        so we can use same session between API calls
+        """
+        if not self.session:
+            self.session = requests.Session()
+
+        m = getattr(self.session, method, None)
+        if not m:
+            raise ValueError(f"method {method} is not supported")
+
+        # pylint: disable=not-callable
+        return m(url, params=params, headers=headers, timeout=self.timeout)
+
+    def close_session(self):
+        if self.session:
+            self.session.close()
+            self.session = None
+
+    def __del__(self):
+        self.close_session()

build_analytics/build_analytics/const.py

@@ -3,7 +3,7 @@
 from enum import IntEnum
 
 # supported schema version
-DB_SCHEMA_VER = 3
+DB_SCHEMA_VER = 4
 
 
 # ENUMS
@@ -13,6 +13,8 @@ class ArchEnum(IntEnum):
     aarch64 = 2
     ppc64le = 3
     s390x = 4
+    src = 5
+    x86_64_v2 = 6
 
 
 class BuildTaskEnum(IntEnum):

build_analytics/build_analytics/extractor/extractor.py

@@ -1,8 +1,10 @@
 # pylint: disable=relative-beyond-top-level
 
+from datetime import datetime, timedelta
 import logging
 from typing import Dict, List
 
+
 from ..api_client import APIclient
 from ..const import BuildTaskEnum
 from ..db import DB
@@ -26,11 +28,13 @@ class Extractor:
         stop = False
 
         while not stop:
+            oldest_build_age = datetime.now().astimezone() - \
+                timedelta(days=self.config.data_store_days)
             logging.info("page: %s", page_num)
             for build in self.api.get_builds(page_num):
                 # check if we shoud stop processing build
                 if build.id <= last_build_id or \
-                   build.created_at <= self.config.oldest_build_age:
+                   build.created_at <= oldest_build_age:
                     stop = True
                     break
 
@@ -73,9 +77,10 @@ class Extractor:
         return build_count
 
     def build_cleanup(self):
-        logging.info('Removing all buidls older then %s',
-                     self.config.oldest_build_age.strftime("%m/%d/%Y, %H:%M:%S"))
-        removed_count = self.db.cleanup_builds(self.config.oldest_build_age)
+        oldest_to_keep = datetime.now().astimezone() - \
+            timedelta(days=self.config.data_store_days)
+        logging.info('Removing all buidls older then %s', oldest_to_keep)
+        removed_count = self.db.cleanup_builds(oldest_to_keep)
         logging.info('removed %d entries', removed_count)
 
     def __update_build_tasks(self, build_tasks: List[BuildTask],
@@ -105,10 +110,11 @@ class Extractor:
                     b.build_id, b.id, BuildTaskEnum(b.status_id).name)
 
     def update_builds(self):
+        not_before = datetime.now().astimezone() - \
+            timedelta(days=self.config.oldest_to_update_days)
         logging.info('Getting unfinished builds that were created after %s ',
-                     self.config.oldest_to_update)
-        unfinished_tasks = self.db.get_unfinished_builds(
-            self.config.oldest_to_update)
+                     not_before)
+        unfinished_tasks = self.db.get_unfinished_builds(not_before)
         for build_id, build_tasks_db in unfinished_tasks.items():
             try:
                 logging.info('Getting status of build %d', build_id)
@@ -137,10 +143,12 @@ class Extractor:
                               build_id, err, exc_info=True)
 
     def updating_test_tasks(self):
+        not_before = datetime.now().astimezone() - \
+            timedelta(days=self.config.oldest_to_update_days)
         logging.info('getting build tasks for builds created after %s',
-                     self.config.oldest_to_update)
+                     not_before)
         build_task_ids = self.db.get_build_tasks_for_tests_update(
-            self.config.oldest_to_update)
+            not_before)
         for build_task_id in build_task_ids:
             try:
                 logging.info('getting tests for build task %s', build_task_id)

build_analytics/build_analytics/extractor/start.py

@@ -1,4 +1,3 @@
-from datetime import datetime, timedelta
 import logging
 from logging.handlers import RotatingFileHandler
 import sys
@@ -23,10 +22,6 @@ def __get_config(yml_path: str) -> ExtractorConfig:
     with open(yml_path, 'r', encoding='utf-8') as flr:
         raw = yaml.safe_load(flr)
 
-    # adding new attrs
-    raw['oldest_build_age'] = datetime.now().astimezone() \
-        - timedelta(days=raw['data_store_days'])
-
     # Dbconfig
     db_params: Dict[str, Any] = {'name': raw['db_name'],
                                  'username': raw['db_username'],
@@ -37,10 +32,6 @@ def __get_config(yml_path: str) -> ExtractorConfig:
         db_params['host'] = raw['db_host']
     raw['db_config'] = DbConfig(**db_params)
 
-    if 'oldest_to_update_days' in raw:
-        raw['oldest_to_update_days'] = datetime.now().astimezone() \
-            - timedelta(days=raw['oldest_to_update_days'])
-
     return ExtractorConfig(**raw)
 
 
@@ -108,7 +99,10 @@ def start(yml_path: str):
         else:
             logging.info('test tasks were updated')
 
+        # freeing up resources
         extractor.db.close_conn()
+        extractor.api.close_session()
+
         logging.info("Extraction was finished")
         logging.info("Sleeping for %d seconds", config.scrape_interval)
         time.sleep(config.scrape_interval)

build_analytics/build_analytics/models/extractor_config.py

@@ -1,4 +1,3 @@
-from datetime import datetime, timedelta
 from pathlib import Path
 
 from pydantic import HttpUrl, Field, BaseModel  # pylint: disable=no-name-in-module
@@ -8,10 +7,10 @@ from .db_config import DbConfig
 # DEFAULTS
 ALBS_URL_DEFAULT = 'https://build.almalinux.org'
 LOG_FILE_DEFAULT = '/tmp/extractor.log'
-API_DEFAULT = 30
+API_TIMEOUT_DEFAULT = 30
 SCRAPE_INTERVAL_DEFAULT = 3600
 START_FROM_DEFAULT = 5808
-OLDEST_TO_UPDATE_DEFAULT = datetime.now().astimezone() - timedelta(days=7)
+OLDEST_TO_UPDATE_DAYS_DEFAULT = 7
 
 
 class ExtractorConfig(BaseModel):
@@ -22,17 +21,17 @@ class ExtractorConfig(BaseModel):
                            default=LOG_FILE_DEFAULT)
     albs_url: HttpUrl = Field(description='ALBS root URL',
                               default=ALBS_URL_DEFAULT)
-    oldest_build_age: datetime = \
-        Field(description='oldest build age to store')
+    data_store_days: int = \
+        Field(description='oldest build (in days) to keep in DB')
     jwt: str = Field(description='ALBS JWT token')
     db_config: DbConfig = Field(description="database configuration")
     api_timeout: int = Field(
         description="max time in seconds to wait for API response",
-        default=API_DEFAULT)
+        default=API_TIMEOUT_DEFAULT)
     scrape_interval: int = Field(description='how often (in seconds) we will extract data from ALBS',
                                  default=SCRAPE_INTERVAL_DEFAULT)
     start_from: int = Field(description='build id to start populating empty db with',
                             default=START_FROM_DEFAULT)
-    oldest_to_update: datetime = \
-        Field(description='oldest unfinished object (build/task/step...) that we will try to update',
-              default=OLDEST_TO_UPDATE_DEFAULT)
+    oldest_to_update_days: int = \
+        Field(description='oldest (in days) unfinished object (build/task/step...) that we will try to update',
+              default=OLDEST_TO_UPDATE_DAYS_DEFAULT)

build_analytics/config_default.yml

@@ -10,7 +10,6 @@ albs_url: https://build.almalinux.org
 # required: yes
 jwt: ""
 
-
 # db_host
 # IP/hostname of database server
 # required: no
@@ -28,7 +27,6 @@ db_port: 5432
 # required: yes
 db_username: albs_analytics
 
-
 # db_password
 # password to connect with
 # required: yes
@@ -39,7 +37,6 @@ db_password: super_secret_password
 # required: yes
 db_name: albs_analytics
 
-
 # log_file
 # file to  write logs to
 # required: no
@@ -62,7 +59,7 @@ scrape_interval: 3600
 # default: 5808 (first build with correct metrics)
 start_from: 5808
 
-# oldest_to_update
+# oldest_to_update_days
 # oldest (in days) unfinished object (build/task/step...) that we will try to update
 # required: false
 # default: 7

build_analytics/migrations/4.sql

@@ -0,0 +1,11 @@
+BEGIN;
+
+INSERT INTO arch_enum (id, value)
+VALUES
+    (5, 'src'),
+    (6, 'x86_64_v2');
+
+UPDATE schema_version
+SET version = 4;
+
+COMMIT;

releases.txt

@@ -27,6 +27,18 @@ build-analytics
     - [ALBS-1077] start deleting builds that were removed from ALBS
   Bugfixes
     - 'Key error' when db_port/db_host is not set
-    - update_builds() ignoring opldest_to_update attribute
+    - update_builds() ignoring odldest_to_update attribute
     - [ALBS-1099] Test task started_at attribute is NULL
     - Max recursion error in 'Test task details.json'
+
+0.3.4 (2023-05-12)
+build_analytics
+  - Bigfix ALBS-1111
+
+0.3.5 (2023-06-01)
+build_analytics:
+  ALBS-1103 start using persistent HTTP connections
+
+0.3.6 (2024-10-08)
+build_analytics: 
+  buildsystem#360 Added src and x86_64_v2 arches