jumpserver/apps/terminal/utils.py

244 lines
9.1 KiB
Python

# -*- coding: utf-8 -*-
#
import os
from itertools import groupby
from django.conf import settings
from django.core.files.storage import default_storage
from django.utils.translation import ugettext as _
import jms_storage
from common.tasks import send_mail_async
from common.utils import get_logger, reverse
from . import const
from .models import ReplayStorage, Session, Command
logger = get_logger(__name__)
def find_session_replay_local(session):
# 新版本和老版本的文件后缀不同
session_path = session.get_rel_replay_path() # 存在外部存储上的路径
local_path = session.get_local_path()
local_path_v1 = session.get_local_path(version=1)
# 去default storage中查找
for _local_path in (local_path, local_path_v1, session_path):
if default_storage.exists(_local_path):
url = default_storage.url(_local_path)
return _local_path, url
return None, None
def download_session_replay(session):
session_path = session.get_rel_replay_path() # 存在外部存储上的路径
local_path = session.get_local_path()
replay_storages = ReplayStorage.objects.all()
configs = {
storage.name: storage.config
for storage in replay_storages
if not storage.type_null_or_server
}
if settings.SERVER_REPLAY_STORAGE:
configs['SERVER_REPLAY_STORAGE'] = settings.SERVER_REPLAY_STORAGE
if not configs:
msg = "Not found replay file, and not remote storage set"
return None, msg
# 保存到storage的路径
target_path = os.path.join(default_storage.base_location, local_path)
target_dir = os.path.dirname(target_path)
if not os.path.isdir(target_dir):
os.makedirs(target_dir, exist_ok=True)
storage = jms_storage.get_multi_object_storage(configs)
ok, err = storage.download(session_path, target_path)
if not ok:
msg = "Failed download replay file: {}".format(err)
logger.error(msg)
return None, msg
url = default_storage.url(local_path)
return local_path, url
def get_session_replay_url(session):
local_path, url = find_session_replay_local(session)
if local_path is None:
local_path, url = download_session_replay(session)
return local_path, url
class ComputeStatUtil:
# system status
@staticmethod
def _common_compute_system_status(value, thresholds):
if thresholds[0] <= value <= thresholds[1]:
return const.ComponentStatusChoices.normal.value
elif thresholds[1] < value <= thresholds[2]:
return const.ComponentStatusChoices.high.value
else:
return const.ComponentStatusChoices.critical.value
@classmethod
def _compute_system_stat_status(cls, stat):
system_stat_thresholds_mapper = {
'cpu_load': [0, 5, 20],
'memory_used': [0, 85, 95],
'disk_used': [0, 80, 99]
}
system_status = {}
for stat_key, thresholds in system_stat_thresholds_mapper.items():
stat_value = getattr(stat, stat_key)
if stat_value is None:
msg = 'stat: {}, stat_key: {}, stat_value: {}'
logger.debug(msg.format(stat, stat_key, stat_value))
stat_value = 0
status = cls._common_compute_system_status(stat_value, thresholds)
system_status[stat_key] = status
return system_status
@classmethod
def compute_component_status(cls, stat):
if not stat:
return const.ComponentStatusChoices.offline
system_status_values = cls._compute_system_stat_status(stat).values()
if const.ComponentStatusChoices.critical in system_status_values:
return const.ComponentStatusChoices.critical
elif const.ComponentStatusChoices.high in system_status_values:
return const.ComponentStatusChoices.high
else:
return const.ComponentStatusChoices.normal
class TypedComponentsStatusMetricsUtil(object):
def __init__(self):
self.components = []
self.grouped_components = []
self.get_components()
def get_components(self):
from .models import Terminal
components = Terminal.objects.filter(is_deleted=False).order_by('type')
grouped_components = groupby(components, lambda c: c.type)
grouped_components = [(i[0], list(i[1])) for i in grouped_components]
self.grouped_components = grouped_components
self.components = components
def get_metrics(self):
metrics = []
for _tp, components in self.grouped_components:
normal_count = high_count = critical_count = 0
total_count = offline_count = session_online_total = 0
for component in components:
total_count += 1
if not component.is_alive:
offline_count += 1
continue
if component.is_normal:
normal_count += 1
elif component.is_high:
high_count += 1
else:
# critical
critical_count += 1
session_online_total += component.get_online_session_count()
metrics.append({
'total': total_count,
'normal': normal_count,
'high': high_count,
'critical': critical_count,
'offline': offline_count,
'session_active': session_online_total,
'type': _tp,
})
return metrics
class ComponentsPrometheusMetricsUtil(TypedComponentsStatusMetricsUtil):
def __init__(self):
super().__init__()
self.metrics = self.get_metrics()
@staticmethod
def convert_status_metrics(metrics):
return {
'any': metrics['total'],
'normal': metrics['normal'],
'high': metrics['high'],
'critical': metrics['critical'],
'offline': metrics['offline']
}
def get_component_status_metrics(self):
prometheus_metrics = list()
# 各组件状态个数汇总
prometheus_metrics.append('# JumpServer 各组件状态个数汇总')
status_metric_text = 'jumpserver_components_status_total{component_type="%s", status="%s"} %s'
for metric in self.metrics:
tp = metric['type']
prometheus_metrics.append(f'## 组件: {tp}')
status_metrics = self.convert_status_metrics(metric)
for status, value in status_metrics.items():
metric_text = status_metric_text % (tp, status, value)
prometheus_metrics.append(metric_text)
return prometheus_metrics
def get_component_session_metrics(self):
prometheus_metrics = list()
# 各组件在线会话数汇总
prometheus_metrics.append('# JumpServer 各组件在线会话数汇总')
session_active_metric_text = 'jumpserver_components_session_active_total{component_type="%s"} %s'
for metric in self.metrics:
tp = metric['type']
prometheus_metrics.append(f'## 组件: {tp}')
metric_text = session_active_metric_text % (tp, metric['session_active'])
prometheus_metrics.append(metric_text)
return prometheus_metrics
def get_component_stat_metrics(self):
prometheus_metrics = list()
# 各组件节点指标
prometheus_metrics.append('# JumpServer 各组件一些指标')
state_metric_text = 'jumpserver_components_%s{component_type="%s", component="%s"} %s'
stats_key = [
'cpu_load', 'memory_used', 'disk_used', 'session_online'
]
old_stats_key = [
'system_cpu_load_1', 'system_memory_used_percent',
'system_disk_used_percent', 'session_active_count'
]
old_stats_key_mapper = dict(zip(stats_key, old_stats_key))
for stat_key in stats_key:
prometheus_metrics.append(f'## 指标: {stat_key}')
for component in self.components:
if not component.is_alive:
continue
component_stat = component.latest_stat
if not component_stat:
continue
metric_text = state_metric_text % (
stat_key, component.type, component.name, getattr(component_stat, stat_key)
)
prometheus_metrics.append(metric_text)
old_stat_key = old_stats_key_mapper.get(stat_key)
old_metric_text = state_metric_text % (
old_stat_key, component.type, component.name, getattr(component_stat, stat_key)
)
prometheus_metrics.append(old_metric_text)
return prometheus_metrics
def get_prometheus_metrics_text(self):
prometheus_metrics = list()
for method in [
self.get_component_status_metrics,
self.get_component_session_metrics,
self.get_component_stat_metrics
]:
prometheus_metrics.extend(method())
prometheus_metrics.append('\n')
prometheus_metrics_text = '\n'.join(prometheus_metrics)
return prometheus_metrics_text