Mini Shell

# coding=utf-8
#
# Copyright © Cloud Linux GmbH & Cloud Linux Software, Inc 2010-2020 All Rights Reserved
#
# Licensed under CLOUD LINUX LICENSE AGREEMENT
# http://cloudlinux.com/docs/LICENSE.TXT

import logging
import os
import re

from sqlalchemy.exc import SQLAlchemyError

from clcommon.clpwd import ClPwd
from lvestats.core.plugin import LveStatsPlugin
from lvestats.lib.commons.func import reboot_lock
from lvestats.lib.commons.sizeutil import dbgov_io_bytes_value
from lvestats.orm.history_gov import history_gov

MAX_FILES_PER_TRANSACTION = 1000


class DBGovSaver(LveStatsPlugin):
    DBSTAT_DIR = "/var/lve/dbgovernor/"
    FILE_PATTERN = re.compile(r"governor\.[0-9]+$", re.IGNORECASE)
    _history_gov_col = list(history_gov.__table__.columns.keys())

    def __init__(self):
        self.logger = logging.getLogger(__name__)
        self.server_id = 'localhost'
        self.engine = None
        self.headers = (
            ('username', str),
            None,  # max_simultaneous_requests not support
            ('sum_cpu', float),
            ('sum_write', float),
            ('sum_read', float),
            None,  # max_cpu not support
            None,  # max_write not support
            None,  # max_read not support
            ('number_of_restricts', int),
            ('limit_cpu_on_period_end', int),
            ('limit_read_on_period_end', int),
            ('limit_write_on_period_end', int),
            ('cause_of_restrict', int),
            ('uid', int),
        )
        self._headers_len = len(self.headers)
        self.cl_pwd = ClPwd()
        self.min_uid = self.cl_pwd.get_sys_min_uid(500)

    def set_config(self, config):
        self.server_id = config.get('server_id', self.server_id)

    def get_user_id(self, username):
        try:
            return self.cl_pwd.get_uid(username)
        except self.cl_pwd.NoSuchUserException as e:
            self.logger.debug('Can not obtain user id for "%s"; %s', username, e)
            return -1

    def scan_dir(self):
        """
        Scans directory generated by db governer and prepares statistics for insertion into database.
        :return: list of tuples [(file name, [lines]), (file name, [lines])...]
        """
        if os.path.exists(self.DBSTAT_DIR):
            flist = filter(self.FILE_PATTERN.search, os.listdir(self.DBSTAT_DIR))
            for f in flist:
                try:
                    file_name = os.path.join(self.DBSTAT_DIR, f)
                    with open(file_name, 'r', encoding='utf-8') as f_stats:
                        f_stats_lines = f_stats.readlines()
                    yield file_name, f_stats_lines
                except IOError:
                    self.logger.warning("No file statistic")
                except UnicodeDecodeError:
                    with open(file_name, 'r', errors='surrogateescape', encoding='utf-8') as file:
                        f_source = file.read()
                        self.logger.error(
                            'Error while decoding the file %s',
                            f,
                            exc_info=True,
                            extra={f: f_source},
                        )
                        yield file_name, []

    def write_to_db(self, conn, scanned):
        """
        :type scanned: generator
        :type conn: sqlalchemy.engine.base.Connection
        :rtype: list(dict(str, int|str))
        """
        values_list = []
        unlink_list = []
        for n_, (file_name, lines) in enumerate(scanned):
            if MAX_FILES_PER_TRANSACTION < n_:
                break
            for line in lines:
                try:
                    self.logger.debug("write: %s", line)
                    line_splited = line.strip().split(';')
                    file_timestamp = int(file_name.split('.')[-1])
                    values = {'server_id': self.server_id, 'ts': file_timestamp}
                    values.update(dict([(h_[0], h_[1](v_)) for h_, v_ in zip(self.headers, line_splited) if h_]))
                    if not values.get('uid'):  # for backward compatibility with governor-mysql < 1.2-1
                        values['uid'] = self.get_user_id(values['username'])  # extend dict by user id
                    if values['uid'] >= self.min_uid:  # ignoring system users and when we can't extract user id
                        values_list.append(values)
                except (IndexError, ValueError):
                    self.logger.warning('Can not parse file %s; data from file not be writen to database', file_name)
            unlink_list.append(file_name)

        # Data to transfer to CM plugin
        data_for_cm = {}

        # insert all data per one commit
        with reboot_lock():
            if values_list:
                try:
                    # filter for insert only supported columns
                    values_list_filtered = [
                        {k: v for k, v in list(d.items()) if k in self._history_gov_col} for d in values_list
                    ]
                    # form data for CM plugin
                    for dbgov_data in values_list_filtered:
                        uid = dbgov_data['uid']
                        data_for_cm[uid] = {
                            'cpu_limit': dbgov_data['limit_cpu_on_period_end'],
                            'io_limit': dbgov_io_bytes_value(
                                dbgov_data['limit_read_on_period_end'], dbgov_data['limit_write_on_period_end']
                            ),
                            'cpu_usage': round(dbgov_data['sum_cpu'], 1),
                            'io_usage': dbgov_io_bytes_value(dbgov_data['sum_read'], dbgov_data['sum_write']),
                        }
                    conn.execute(history_gov.__table__.insert(), values_list_filtered)
                except (SQLAlchemyError, KeyError) as e:
                    self.logger.warning(str(e))
            try:
                list(map(os.unlink, unlink_list))
            except OSError:
                pass

        return data_for_cm

    def execute(self, lve_data):
        """
        :type lve_data: dict
        """
        if "dbgov_data" not in lve_data:
            lve_data["dbgov_data"] = []
        conn = self.engine.connect()
        try:
            scanned = self.scan_dir()
            dbgov_data_for_cm = self.write_to_db(conn, scanned)
            if dbgov_data_for_cm:
                lve_data["dbgov_data_for_cm"] = dbgov_data_for_cm
        finally:
            conn.close()