𝗖𝗨𝗥𝗥𝗘𝗡𝗧 𝗙𝗜𝗟𝗘 : /opt/cloudlinux/venv/lib/python3.11/site-packages/cl_website_collector//constants.py
# -*- coding: utf-8 -*-
#
# Copyright © Cloud Linux GmbH & Cloud Linux Software, Inc 2010-2024 All Rights Reserved
#
# Licensed under CLOUD LINUX LICENSE AGREEMENT
# http://cloudlinux.com/docs/LICENSE.TXT
from pathlib import Path
# Collector version
# once version is increased - collector invalidates cache
# see sending_stats dict in website_collector.py
MAJOR_VER = 2
VERSION = f"{MAJOR_VER}.0.0"
# Storage dir
STORAGE_DIR = Path("/var/lve/website_collector")
# Opt-out control
OPT_OUT_FILE = STORAGE_DIR / ".disabled"
# Processing limits
DOCROOT_SCAN_TIMEOUT = 30 # Timeout per docroot scan in seconds
DOCROOT_LIMIT_PER_RUN = 500 # Maximum number of docroots to process in single run
DOCROOT_MAX_DEPTH = 30 # Maximum depth of docroot scanning
DOCROOT_EXCLUDE_DIRS = [
'node_modules', 'vendor',
'.idea', '.vscode', '.well-known',
'.git', '.svn', '.hg'] # Directories to exclude from docroot scanning
SENDING_TIMEOUT = 7200 # Total timeout 2 hours
SENDING_COMPRESSION = True # Compress data before sending
# MAX_RECORD_SIZE: JSON-payload batch limit before HTTP split; ~10 files
# per request keeps batching while bounding payload size.
MAX_RECORD_SIZE = 1024 * 1024 * 10 # 10MB
# FILE_SIZE_LIMIT: per-file read cap; blocks tenant-user DoS via an
# oversized .htaccess (CLOS-4275 #17).
FILE_SIZE_LIMIT = 1024 * 1024 # 1MB
