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

@@ -1,7 +1,7 @@
 from datetime import datetime
 import logging
 from urllib.parse import urljoin
-from typing import Dict, List, Any
+from typing import Dict, List, Any, Optional
 
 import requests
 
@@ -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 = []
@@ -47,11 +48,18 @@ class APIclient():
                               b, err, exc_info=True)
         return result
 
-    def get_build(self, build_id: int) -> Build:
+    def get_build(self, build_id: int) -> Optional[Build]:
+        '''
+        method returns None if build was deleted from ALBS
+        '''
         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
+
         response.raise_for_status()
         return self._parse_build(response.json())
 
@@ -104,7 +112,8 @@ class APIclient():
                 if not task['performance_stats']:
                     logging.warning(
                         "no perfomance_stats for build_id: %s, build_task_id: %s", build_id, task['id'])
-                    stats = {'build_node_stats': {}, 'build_done_stats': {}}
+                    stats: dict[str, Any] = {
+                        'build_node_stats': {}, 'build_done_stats': {}}
                 else:
                     stats = task['performance_stats'][0]['statistics']
 
@@ -161,17 +170,10 @@ class APIclient():
         return result
 
     def __parse_test_tasks(self, raw_tasks: List[Dict[str, Any]],
-                           build_task_id: int,
-                           started_at: str = None) -> List[TestTask]:
+                           build_task_id: int) -> List[TestTask]:
         result: List[TestTask] = []
         for task in raw_tasks:
             if task['alts_response']:
-                try:
-                    started_raw = task['alts_response']['stats']['started_at']
-                except KeyError:
-                    started_at = None
-                else:
-                    started_at = datetime.fromisoformat(started_raw+TZ_OFFSET)
                 try:
                     stats_raw = task['alts_response']['stats']
                 except KeyError:
@@ -179,7 +181,6 @@ class APIclient():
                 else:
                     steps_stats = self.__parse_test_steps_stats(stats_raw)
             else:
-                started_at = None
                 steps_stats = None
             params = {
                 'id': task['id'],
@@ -189,7 +190,8 @@ class APIclient():
                 'package_fullname': '_'.join([task['package_name'],
                                               task['package_version'],
                                               task['package_release']]),
-                'started_at': started_at,
+                'started_at': self.__get_test_task_started_at(
+                    steps_stats) if steps_stats else None,
                 'steps_stats': steps_stats
             }
 
@@ -209,3 +211,49 @@ class APIclient():
                     p[k] = datetime.fromisoformat(p[k]+TZ_OFFSET)
             teast_steps_params[field_name] = TestStepStat(**p)
         return TestStepsStats(**teast_steps_params)
+
+    def __get_test_task_started_at(self, stats: TestStepsStats) -> Optional[datetime]:
+        """
+        getting started_at attribute for test by using oldest start_ts timestamp
+        among all test tasks steps
+        """
+        if not stats:
+            return None
+
+        start_ts = None
+        for field_name in stats.__fields__.keys():
+            stat: TestStepStat = getattr(stats, field_name)
+            if not stat:
+                continue
+            if not start_ts or start_ts > stat.start_ts:
+                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/db.py

@@ -62,34 +62,34 @@ class DB():
                           build_task.started_at, build_task.finished_at, build_task.status_id))
 
         # inserting web node stats
-        for stat in web_node_stats:
+        for wn_stat in web_node_stats:
 
             # do not insert empty stats
-            if stat.start_ts is None:
+            if wn_stat.start_ts is None:
                 continue
 
             sql = '''
                     INSERT INTO web_node_stats (build_task_id, stat_name_id, start_ts, end_ts)
                     VALUES (%s, %s, %s, %s);
                 '''
-            cur.execute(sql, (stat.build_task_id, stat.stat_name_id,
-                        stat.start_ts, stat.end_ts))
+            cur.execute(sql, (wn_stat.build_task_id, wn_stat.stat_name_id,
+                        wn_stat.start_ts, wn_stat.end_ts))
             logging.debug('raw SQL query: %s', cur.query)
             self.__conn.commit()
 
         # inserting build node stats
-        for stat in build_node_stats:
+        for bn_stat in build_node_stats:
 
             # do not insert empty stats
-            if stat.start_ts is None:
+            if bn_stat.start_ts is None:
                 continue
 
             sql = '''
                     INSERT INTO build_node_stats(build_task_id, stat_name_id, start_ts, end_ts)
                     VALUES (%s, %s, %s, %s);
                 '''
-            cur.execute(sql, (stat.build_task_id, stat.stat_name_id,
-                        stat.start_ts, stat.end_ts))
+            cur.execute(sql, (bn_stat.build_task_id, bn_stat.stat_name_id,
+                        bn_stat.start_ts, bn_stat.end_ts))
             logging.debug('raw SQL query: %s', cur.query)
 
         # commiting changes
@@ -121,11 +121,12 @@ class DB():
 
         # getting unfinished builds
         sql = 'SELECT id FROM builds where finished_at is NULL AND created_at > %s;'
+        builds_to_check: Dict[int, bool] = {}
         cur = self.__conn.cursor()
         cur.execute(sql, (not_before.timestamp(),))
         logging.debug('raw SQL query: %s', cur.query)
         for row in cur.fetchall():
-            res[row[0]] = {}
+            builds_to_check[row[0]] = True
 
         # getting list of unfinished tasks
         sql = 'SELECT id, build_id, status_id FROM build_tasks WHERE status_id < 2;'
@@ -135,6 +136,8 @@ class DB():
             build_task_id: int = row[0]
             build_id: int = row[1]
             status_id: int = row[2]
+            if build_id not in builds_to_check:
+                continue
             try:
                 res[build_id][build_task_id] = status_id
             except KeyError:
@@ -195,11 +198,11 @@ class DB():
             logging.debug('raw SQL query: %s', cur.query)
 
         # updating build_node_stats
-        for stat in build_node_stats:
+        for bn_stat in build_node_stats:
             logging.debug(
-                'updating build_node_stats %s build_task %s', stat.stat_name_id, build_task.id)
-            if self.stat_exists(task_id=stat.build_task_id,
-                                stat_name_id=stat.stat_name_id,
+                'updating build_node_stats %s build_task %s', bn_stat.stat_name_id, build_task.id)
+            if self.stat_exists(task_id=bn_stat.build_task_id,
+                                stat_name_id=bn_stat.stat_name_id,
                                 table_name='build_node_stats',
                                 column_name='build_task_id'):
                 sql = '''
@@ -213,9 +216,9 @@ class DB():
                         VALUES (%(build_task_id)s, %(stat_name_id)s, %(start_ts)s, %(end_ts)s);
                       '''
             params = {'build_task_id': build_task.id,
-                      'stat_name_id': stat.stat_name_id,
-                      'start_ts': stat.start_ts,
-                      'end_ts': stat.end_ts}
+                      'stat_name_id': bn_stat.stat_name_id,
+                      'start_ts': bn_stat.start_ts,
+                      'end_ts': bn_stat.end_ts}
             logging.debug('raw SQL query: %s', cur.query)
             cur.execute(sql, params)
 
@@ -318,3 +321,11 @@ class DB():
                                 s.start_ts, s.finish_ts))
         # commiting changes
         self.__conn.commit()
+
+    def delete_build(self, build_id: int):
+        params = (build_id,)
+        sql = "DELETE FROM builds WHERE id = %s;"
+        cur = self.__conn.cursor()
+
+        cur.execute(sql, params)
+        self.__conn.commit()

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],
@@ -105,13 +110,20 @@ class Extractor:
                     b.build_id, b.id, BuildTaskEnum(b.status_id).name)
 
     def update_builds(self):
-        logging.info('Getting list of tasks from DB')
-        unfinished_tasks = self.db.get_unfinished_builds(
-            self.config.oldest_to_update)
+        not_before = datetime.now().astimezone() - \
+            timedelta(days=self.config.oldest_to_update_days)
+        logging.info('Getting unfinished builds that were created after %s ',
+                     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)
                 build = self.api.get_build(build_id)
+                if not build:
+                    logging.warning(
+                        "build %s was deleted from albs, removing it", build_id)
+                    self.db.delete_build(build_id)
+                    continue
 
                 logging.info('Updating build tasks')
                 build_tasks_to_check = [
@@ -131,10 +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

@@ -1,8 +1,8 @@
-from datetime import datetime, timedelta
 import logging
 from logging.handlers import RotatingFileHandler
 import sys
 import time
+from typing import Dict, Any
 
 import yaml
 
@@ -22,19 +22,15 @@ 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'])
-
-    raw['db_config'] = DbConfig(name=raw['db_name'],
-                                port=int(raw['db_port']),
-                                host=raw['db_host'],
-                                username=raw['db_username'],
-                                password=raw['db_password'])
-
-    if 'oldest_to_update_days' in raw:
-        raw['oldest_to_update_days'] = datetime.now().astimezone() \
-            - timedelta(days=raw['oldest_to_update_days'])
+    # Dbconfig
+    db_params: Dict[str, Any] = {'name': raw['db_name'],
+                                 'username': raw['db_username'],
+                                 'password': raw['db_password'], }
+    if 'db_port' in raw:
+        db_params['port'] = raw['db_port']
+    if 'db_host' in raw:
+        db_params['host'] = raw['db_host']
+    raw['db_config'] = DbConfig(**db_params)
 
     return ExtractorConfig(**raw)
 
@@ -103,7 +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/db_config.py

@@ -1,9 +1,13 @@
 from pydantic import BaseModel, Field
 
 
+DB_PORT = 5432
+DB_HOST = "localhost"
+
+
 class DbConfig(BaseModel):
     name: str = Field(description="db name")
-    port: int = Field(description="db server port")
-    host: str = Field(description="db server ip/hostname")
+    port: int = Field(description="db server port", default=DB_PORT)
+    host: str = Field(description="db server ip/hostname", default=DB_HOST)
     username: str = Field(description="username to connect with")
     password: str = Field(description="password to connect with1")

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
@@ -60,9 +57,9 @@ scrape_interval: 3600
 # build_id to start populating empty db with
 # required: false
 # default: 5808 (first build with correct metrics)
-start_from:
+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/grafana-dashbords/Test task details.json

@@ -1,8 +1,8 @@
 {
   "__inputs": [
     {
-      "name": "DS_POSTGRESQL",
-      "label": "PostgreSQL",
+      "name": "DS_ALBS_ANALYTICS",
+      "label": "albs_analytics",
       "description": "",
       "type": "datasource",
       "pluginId": "postgres",
@@ -21,7 +21,7 @@
       "type": "grafana",
       "id": "grafana",
       "name": "Grafana",
-      "version": "9.3.2"
+      "version": "9.3.6"
     },
     {
       "type": "datasource",
@@ -58,7 +58,7 @@
       }
     ]
   },
-  "editable": false,
+  "editable": true,
   "fiscalYearStartMonth": 0,
   "graphTooltip": 0,
   "id": null,
@@ -68,7 +68,7 @@
     {
       "datasource": {
         "type": "postgres",
-        "uid": "${DS_POSTGRESQL}"
+        "uid": "${DS_ALBS_ANALYTICS}"
       },
       "fieldConfig": {
         "defaults": {
@@ -152,13 +152,13 @@
         "showHeader": true,
         "sortBy": []
       },
-      "pluginVersion": "9.3.2",
+      "pluginVersion": "9.3.6",
       "targets": [
         {
           "cacheDurationSeconds": 300,
           "datasource": {
             "type": "postgres",
-            "uid": "${DS_POSTGRESQL}"
+            "uid": "${DS_ALBS_ANALYTICS}"
           },
           "editorMode": "code",
           "fields": [
@@ -216,7 +216,7 @@
     {
       "datasource": {
         "type": "postgres",
-        "uid": "${DS_POSTGRESQL}"
+        "uid": "${DS_ALBS_ANALYTICS}"
       },
       "fieldConfig": {
         "defaults": {
@@ -263,13 +263,13 @@
         },
         "showUnfilled": false
       },
-      "pluginVersion": "9.3.2",
+      "pluginVersion": "9.3.6",
       "targets": [
         {
           "cacheDurationSeconds": 300,
           "datasource": {
             "type": "postgres",
-            "uid": "${DS_POSTGRESQL}"
+            "uid": "${DS_ALBS_ANALYTICS}"
           },
           "editorMode": "code",
           "fields": [
@@ -315,7 +315,7 @@
     {
       "datasource": {
         "type": "postgres",
-        "uid": "${DS_POSTGRESQL}"
+        "uid": "${DS_ALBS_ANALYTICS}"
       },
       "fieldConfig": {
         "defaults": {
@@ -409,18 +409,18 @@
         "showHeader": true,
         "sortBy": [
           {
-            "desc": true,
-            "displayName": "finished"
+            "desc": false,
+            "displayName": "started"
           }
         ]
       },
-      "pluginVersion": "9.3.2",
+      "pluginVersion": "9.3.6",
       "targets": [
         {
           "cacheDurationSeconds": 300,
           "datasource": {
             "type": "postgres",
-            "uid": "${DS_POSTGRESQL}"
+            "uid": "${DS_ALBS_ANALYTICS}"
           },
           "editorMode": "code",
           "fields": [
@@ -485,16 +485,16 @@
         "current": {},
         "datasource": {
           "type": "postgres",
-          "uid": "${DS_POSTGRESQL}"
+          "uid": "${DS_ALBS_ANALYTICS}"
         },
-        "definition": "SELECT id\nFROM test_tasks\nORDER BY id DESC",
+        "definition": "SELECT id\nFROM test_tasks\nORDER BY id DESC\nLIMIT 1000;",
         "hide": 0,
         "includeAll": false,
         "label": "Test task id",
         "multi": false,
         "name": "id",
         "options": [],
-        "query": "SELECT id\nFROM test_tasks\nORDER BY id DESC",
+        "query": "SELECT id\nFROM test_tasks\nORDER BY id DESC\nLIMIT 1000;",
         "refresh": 1,
         "regex": "",
         "skipUrlSync": false,
@@ -511,6 +511,6 @@
   "timezone": "",
   "title": "Test task details",
   "uid": "8nFXlkB4z",
-  "version": 8,
+  "version": 3,
   "weekStart": ""
 }

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;

build_analytics/releases.txt

@@ -1,22 +0,0 @@
-0.1.0 (2023-03-01)
-First version
-
-0.2.0 (2023-03-15)
-- New parameter start_from
-- Moved to double persition for timestamps
-- Added metrics for build steps
-
-0.2.1 (2023-03-15)
- - Added canceled Build task status
-
-0.3.0 (2023-03-22)
- - Added test tasks stats
- - New config parameter: oldest_to_update_days
-
-0.3.1 (2023-03-22)
- - db: bugfix with migration to version 3
- - added info about api_timeout config parameter
- - bugfix with processing of test tasks with new revision
-
-0.3.2 (2023-03-23)
- - Bugfix ALBS-1060

releases.txt

@@ -0,0 +1,44 @@
+0.1.0 (2023-03-01)
+First version
+
+0.2.0 (2023-03-15)
+- New parameter start_from
+- Moved to double persition for timestamps
+- Added metrics for build steps
+
+0.2.1 (2023-03-15)
+ - Added canceled Build task status
+
+0.3.0 (2023-03-22)
+ - Added test tasks stats
+ - New config parameter: oldest_to_update_days
+
+0.3.1 (2023-03-22)
+ - db: bugfix with migration to version 3
+ - added info about api_timeout config parameter
+ - bugfix with processing of test tasks with new revision
+
+0.3.2 (2023-03-23)
+ - Bugfix ALBS-1060
+
+0.3.3 (2023-04-24)
+build-analytics
+  Improvements
+    - [ALBS-1077] start deleting builds that were removed from ALBS
+  Bugfixes
+    - 'Key error' when db_port/db_host is not set
+    - 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