db_cleanup plugin

This commit is contained in:
Jokob-sk
2023-09-17 21:46:05 +10:00
parent 80de181827
commit a5b952f18c
13 changed files with 370 additions and 119 deletions

View File

@@ -0,0 +1,7 @@
## Overview
Plugin to run regular database cleanup tasks. It is strongly recommended to have an hourly or at least daily schedule running.
### Usage
- Check the Settings page for details.

View File

@@ -0,0 +1,180 @@
{
"code_name": "db_cleanup",
"unique_prefix": "DBCLNP",
"enabled": true,
"data_source": "script",
"show_ui": false,
"localized": ["display_name", "description", "icon"],
"display_name": [
{
"language_code": "en_us",
"string": "DB cleanup"
}
],
"icon": [
{
"language_code": "en_us",
"string": "<i class=\"fa-solid fa-broom\"></i>"
}
],
"description": [
{
"language_code": "en_us",
"string": "A plugin to schedule database cleanup & upkeep tasks."
}
],
"params" : [{
"name" : "pluginskeephistory",
"type" : "setting",
"value" : "PLUGINS_KEEP_HIST"
},
{
"name" : "daystokeepevents",
"type" : "setting",
"value" : "DAYS_TO_KEEP_EVENTS"
},
{
"name" : "hourstokeepnewdevice",
"type" : "setting",
"value" : "HRS_TO_KEEP_NEWDEV"
},
{
"name" : "pholuskeepdays",
"type" : "setting",
"value" : "PHOLUS_DAYS_DATA"
}
],
"settings": [
{
"function": "RUN",
"type": "text.select",
"default_value":"schedule",
"options": ["disabled", "once", "schedule", "always_after_scan"],
"localized": ["name", "description"],
"name" :[{
"language_code":"en_us",
"string" : "When to run"
},
{
"language_code":"es_es",
"string" : "Cuándo ejecutar"
},
{
"language_code":"de_de",
"string" : "Wann laufen"
}],
"description": [{
"language_code":"en_us",
"string" : "When the cleanup should be performed. An hourly or daily <code>SCHEDULE</code> is a good option."
}]
},
{
"function": "CMD",
"type": "readonly",
"default_value": "python3 /home/pi/pialert/front/plugins/db_cleanup/script.py pluginskeephistory={pluginskeephistory} hourstokeepnewdevice={hourstokeepnewdevice} daystokeepevents={daystokeepevents} pholuskeepdays={pholuskeepdays}",
"options": [],
"localized": ["name", "description"],
"name": [
{
"language_code": "en_us",
"string": "Command"
},
{
"language_code": "es_es",
"string": "Comando"
},
{
"language_code": "de_de",
"string": "Befehl"
}
],
"description": [
{
"language_code": "en_us",
"string": "Command to run. This can not be changed"
},
{
"language_code": "es_es",
"string": "Comando a ejecutar. Esto no se puede cambiar"
},
{
"language_code": "de_de",
"string": "Befehl zum Ausführen. Dies kann nicht geändert werden"
}
]
},
{
"function": "RUN_SCHD",
"type": "text",
"default_value":"*/30 * * * *",
"options": [],
"localized": ["name", "description"],
"name" : [{
"language_code":"en_us",
"string" : "Schedule"
},
{
"language_code":"es_es",
"string" : "Schedule"
},
{
"language_code":"de_de",
"string" : "Schedule"
}],
"description": [{
"language_code":"en_us",
"string" : "Only enabled if you select <code>schedule</code> in the <a href=\"#CSVBCKP_RUN\"><code>CSVBCKP_RUN</code> setting</a>. Make sure you enter the schedule in the correct cron-like format (e.g. validate at <a href=\"https://crontab.guru/\" target=\"_blank\">crontab.guru</a>). For example entering <code>0 4 * * *</code> will run the scan after 4 am in the <a onclick=\"toggleAllSettings()\" href=\"#TIMEZONE\"><code>TIMEZONE</code> you set above</a>. Will be run NEXT time the time passes."
},
{
"language_code":"es_es",
"string" : "Solo está habilitado si selecciona <code>schedule</code> en la configuración <a href=\"#CSVBCKP_RUN\"><code>CSVBCKP_RUN</code></a>. Asegúrese de ingresar la programación en el formato similar a cron correcto (por ejemplo, valide en <a href=\"https://crontab.guru/\" target=\"_blank\">crontab.guru</a>). Por ejemplo, ingresar <code>0 4 * * *</code> ejecutará el escaneo después de las 4 a.m. en el <a onclick=\"toggleAllSettings()\" href=\"#TIMEZONE\"><code>TIMEZONE</ código> que configuró arriba</a>. Se ejecutará la PRÓXIMA vez que pase el tiempo."
},
{
"language_code":"de_de",
"string" : "Nur aktiviert, wenn Sie <code>schedule</code> in der <a href=\"#CSVBCKP_RUN\"><code>CSVBCKP_RUN</code>-Einstellung</a> auswählen. Stellen Sie sicher, dass Sie den Zeitplan im richtigen Cron-ähnlichen Format eingeben (z. B. validieren unter <a href=\"https://crontab.guru/\" target=\"_blank\">crontab.guru</a>). Wenn Sie beispielsweise <code>0 4 * * *</code> eingeben, wird der Scan nach 4 Uhr morgens in der <a onclick=\"toggleAllSettings()\" href=\"#TIMEZONE\"><code>TIMEZONE</ ausgeführt. Code> den Sie oben festgelegt haben</a>. Wird das NÄCHSTE Mal ausgeführt, wenn die Zeit vergeht."
}]
},
{
"function": "RUN_TIMEOUT",
"type": "integer",
"default_value": 30,
"options": [],
"localized": ["name", "description"],
"name": [
{
"language_code": "en_us",
"string": "Run timeout"
},
{
"language_code": "es_es",
"string": "Tiempo límite de ejecución"
},
{
"language_code": "de_de",
"string": "Zeitüberschreitung"
}
],
"description": [
{
"language_code": "en_us",
"string": "Maximum time in seconds to wait for the script to finish. If this time is exceeded the script is aborted."
},
{
"language_code": "es_es",
"string": "Tiempo máximo en segundos para esperar a que finalice el script. Si se supera este tiempo, el script se cancela."
},
{
"language_code": "de_de",
"string": "Maximale Zeit in Sekunden, die auf den Abschluss des Skripts gewartet werden soll. Bei Überschreitung dieser Zeit wird das Skript abgebrochen."
}
]
}
],
"database_column_definitions":
[
]
}

View File

@@ -0,0 +1,147 @@
#!/usr/bin/env python
# test script by running:
# /home/pi/pialert/front/plugins/db_cleanup/script.py pluginskeephistory=250 hourstokeepnewdevice=48 daystokeepevents=90
import os
import pathlib
import argparse
import sys
import hashlib
import csv
import sqlite3
from io import StringIO
from datetime import datetime
sys.path.append("/home/pi/pialert/front/plugins")
sys.path.append('/home/pi/pialert/pialert')
from plugin_helper import Plugin_Object, Plugin_Objects, decodeBase64
from logger import mylog, append_line_to_file
from helper import timeNowTZ
from const import logPath, pialertPath
CUR_PATH = str(pathlib.Path(__file__).parent.resolve())
LOG_FILE = os.path.join(CUR_PATH, 'script.log')
RESULT_FILE = os.path.join(CUR_PATH, 'last_result.log')
def main():
parser = argparse.ArgumentParser(description='DB cleanup tasks')
parser.add_argument('pluginskeephistory', action="store", help="TBC")
parser.add_argument('hourstokeepnewdevice', action="store", help="TBC")
parser.add_argument('daystokeepevents', action="store", help="TBC")
parser.add_argument('pholuskeepdays', action="store", help="TBC")
values = parser.parse_args()
PLUGINS_KEEP_HIST = values.pluginskeephistory.split('=')[1]
HRS_TO_KEEP_NEWDEV = values.hourstokeepnewdevice.split('=')[1]
DAYS_TO_KEEP_EVENTS = values.daystokeepevents.split('=')[1]
PHOLUS_DAYS_DATA = values.pholuskeepdays.split('=')[1]
mylog('verbose', ['[DBCLNP] In script'])
# Execute cleanup/upkeep
cleanup_database('/home/pi/pialert/db/pialert.db', DAYS_TO_KEEP_EVENTS, PHOLUS_DAYS_DATA, HRS_TO_KEEP_NEWDEV, PLUGINS_KEEP_HIST)
mylog('verbose', ['[DBCLNP] Cleanup complete file '])
return 0
#===============================================================================
# Cleanup / upkeep database
#===============================================================================
def cleanup_database (dbPath, DAYS_TO_KEEP_EVENTS, PHOLUS_DAYS_DATA, HRS_TO_KEEP_NEWDEV, PLUGINS_KEEP_HIST):
"""
Cleaning out old records from the tables that don't need to keep all data.
"""
mylog('verbose', ['[DBCLNP] Upkeep Database:' ])
# Connect to the PiAlert SQLite database
conn = sqlite3.connect(dbPath)
cursor = conn.cursor()
# Cleanup Online History
mylog('verbose', ['[DBCLNP] Online_History: Delete all but keep latest 150 entries'])
cursor.execute ("""DELETE from Online_History where "Index" not in (
SELECT "Index" from Online_History
order by Scan_Date desc limit 150)""")
mylog('verbose', ['[DBCLNP] Optimize Database'])
# Cleanup Events
mylog('verbose', [f'[DBCLNP] Events: Delete all older than {str(DAYS_TO_KEEP_EVENTS)} days (DAYS_TO_KEEP_EVENTS setting)'])
cursor.execute (f"""DELETE FROM Events
WHERE eve_DateTime <= date('now', '-{str(DAYS_TO_KEEP_EVENTS)} day')""")
# Trim Plugins_History entries to less than PLUGINS_KEEP_HIST setting per unique "Plugin" column entry
mylog('verbose', [f'[DBCLNP] Plugins_History: Trim Plugins_History entries to less than {str(PLUGINS_KEEP_HIST)} per Plugin (PLUGINS_KEEP_HIST setting)'])
# Build the SQL query to delete entries that exceed the limit per unique "Plugin" column entry
delete_query = f"""DELETE FROM Plugins_History
WHERE "Index" NOT IN (
SELECT "Index"
FROM (
SELECT "Index",
ROW_NUMBER() OVER(PARTITION BY "Plugin" ORDER BY DateTimeChanged DESC) AS row_num
FROM Plugins_History
) AS ranked_objects
WHERE row_num <= {str(PLUGINS_KEEP_HIST)}
);"""
cursor.execute(delete_query)
# Cleanup Pholus_Scan
if PHOLUS_DAYS_DATA != 0:
mylog('verbose', ['[DBCLNP] Pholus_Scan: Delete all older than ' + str(PHOLUS_DAYS_DATA) + ' days (PHOLUS_DAYS_DATA setting)'])
# todo: improvement possibility: keep at least N per mac
cursor.execute (f"""DELETE FROM Pholus_Scan
WHERE Time <= date('now', '-{str(PHOLUS_DAYS_DATA)} day')""")
# Cleanup New Devices
if HRS_TO_KEEP_NEWDEV != 0:
mylog('verbose', [f'[DBCLNP] Devices: Delete all New Devices older than {str(HRS_TO_KEEP_NEWDEV)} hours (HRS_TO_KEEP_NEWDEV setting)'])
cursor.execute (f"""DELETE FROM Devices
WHERE dev_NewDevice = 1 AND dev_FirstConnection < date('now', '+{str(HRS_TO_KEEP_NEWDEV)} hour')""")
# De-dupe (de-duplicate) from the Plugins_Objects table
# TODO This shouldn't be necessary - probably a concurrency bug somewhere in the code :(
mylog('verbose', ['[DBCLNP] Plugins_Objects: Delete all duplicates'])
cursor.execute("""
DELETE FROM Plugins_Objects
WHERE rowid > (
SELECT MIN(rowid) FROM Plugins_Objects p2
WHERE Plugins_Objects.Plugin = p2.Plugin
AND Plugins_Objects.Object_PrimaryID = p2.Object_PrimaryID
AND Plugins_Objects.Object_SecondaryID = p2.Object_SecondaryID
AND Plugins_Objects.UserData = p2.UserData
)
""")
# De-Dupe (de-duplicate - remove duplicate entries) from the Pholus_Scan table
mylog('verbose', ['[DBCLNP] Pholus_Scan: Delete all duplicates'])
cursor.execute ("""DELETE FROM Pholus_Scan
WHERE rowid > (
SELECT MIN(rowid) FROM Pholus_Scan p2
WHERE Pholus_Scan.MAC = p2.MAC
AND Pholus_Scan.Value = p2.Value
AND Pholus_Scan.Record_Type = p2.Record_Type
);""")
conn.commit()
# Shrink DB
mylog('verbose', ['[DBCLNP] Shrink Database'])
cursor.execute ("VACUUM;")
# Close the database connection
conn.close()
#===============================================================================
# BEGIN
#===============================================================================
if __name__ == '__main__':
main()